Donor-defined mesenchymal stem cell antimicrobial potency against nontuberculous mycobacterium.

Chronic nontuberculous mycobacterial infections with Mycobacterium avium and Mycobacterium intracellulare complicate bronchiectasis, chronic obstructive airway disease, and the health of aging individuals. These insidious intracellular pathogens cause considerable morbidity and eventual mortality in individuals colonized with these bacteria. Current treatment regimens with antibiotic macrolides are both toxic and often inefficient at providing infection resolution. In this article, we demonstrate that human marrow-derived mesenchymal stem cells are antimicrobial and anti-inflammatory in vitro and in the context of an in vivo sustained infection of either M. avium and/or M. intracellulare.

Human mesenchymal stem cells have significant clinical utility in treating scenarios of infection and inflammation. Selection of the mesenchymal stem cell source defines the degree of therapeutic potency and potentially efficacy. The studies outlined in this article define human mesenchymal stem cell potency against intracellular bacterial pathogens that are difficult to treat. Choosing the optimal donor for therapeutic impact may provide better outcomes in clinical trials using mesenchymal stem cells for therapeutic potency against these pathogens.

INTRODUCTION

There are over 100 000 people in the United States who suffer from chronic nontuberculous mycobacterial pulmonary infections. The infections are difficult to treat, and they significantly impair the quality of life and survival of patients. Nontuberculous mycobacteria (NTMs) are opportunistic intracellular bacteria that are environmentally ubiquitous, becoming pathogenic in individuals with compromised immune systems, including patients with cystic fibrosis (CF), bronchiectasis, pulmonary alveolar proteinosis, or chronic obstructive pulmonary disease (COPD). NTM infections are also prevalent in older individuals (aged >60 years at 13.9 cases per 100 000 people); however, the reasons for this susceptibility remain elusive. Treatment for NTM infections includes a 12‐month regimen of a macrolide‐based, multidrug cocktail, which is often associated with toxicity and a significant relapse rate (12%‐58%).

Mycobacterium avium complex (MAC) is a complex of NTM species and is the most common NTM isolate identified in non‐CF bronchiectasis, COPD, and aging individuals. Although the species and relative predominance of the NTM that colonizes patients is different, the main players in MAC are M. avium and Mycobacterium intracellulare. MACs, like M. avium and M. intracellulare, are slow‐growing bacteria that have complicated the development of standardized in vitro and in vivo modeling systems that can be used to monitor new anti‐NTM therapeutics. In previous studies, we developed an innovative protocol in which M. avium and M. intracellulare can be evaluated over the course of 1 week instead of 4 to 6 weeks. This modeling system is used in this article to evaluate the potential of human mesenchymal stem cells (hMSCs) to treat NTM infections.

hMSCs are dynamic storehouses of antimicrobial activity. Our studies have shown that hMSCs have antimicrobial, anti‐inflammatory, and antifibrotic potential both in vitro and in vivo. , , We have also documented that hMSC treatment results in improved effectiveness of antibiotics, thus decreasing the dose required for eradication of bacteria. The studies in this article use in vitro and in vivo models of chronic NTM infection to evaluate the potential therapeutic use of hMSCs. Using these models, we identified donor‐specific hMSC potency against M. avium and M. intracellulare. These data demonstrate the compelling versatility of hMSCs and the potential use of our models to define the ideal therapeutic regimen to treat chronic NTM infections in vivo. Additionally, our studies define the unique hMSC fingerprint related to the capacity to manage NTM infections evaluating the potential for hMSCs to produce heme oxygenase 1 (HMOX‐1), interleukin (IL)‐6, and macrophage inflammatory protein 3α (CCL20), which have the capacity to contribute to nontuberculous mycobacterial infection management. , , ,

MATERIALS AND METHODS

Mice

All procedures involving mice were reviewed and approved by Case Western Reserve University's Institutional Animal Care and Use Committee (IACUC 2014‐0093). Studies used C57BL/6J wild‐type (WT) mice and the CF mouse model B6.129 Cftr tm1Kth Tg(FABPCFTR)1Jaw/Cwr (gut‐corrected delF508) and Cftr tm1Kth (complete delF508 knockout), which has a cystic fibrosis‐like overactive inflammatory response to infection, , , inefficient in resolving pathogens. , The complete delF508 has a severe response to infection with high mortality and requires laxatives to attenuate gut obstruction. These are highly expensive mice and hard to breed. The gut‐corrected version of delF508 retains the pulmonary hyperinflammatory response to infection, without the same level of mortality associated with the gut contributions to disease. For the purpose of profiling hMSC potency, we use the sublethal model for many of the studies, which were then followed up with the total delF508 strain. Both mouse strains are congenic over several generations on a C57BL/6J background. Although the CF mouse does not recapitulate many aspects of CF lung pathophysiology, it has become the first‐line standardized model to monitor infection resolution and uncontrolled inflammation. , , Other models, like the airway‐specific over‐expression of the epithelial Na(+) channel (βENaC) mouse models, provide alternative pathophysiology testing endpoints such as mucociliary clearance or, like the ferret and pig models, have more complete representation of CF pathophysiology. , Our studies use the murine CF modeling system because they are the first line of studies to evaluate anti‐inflammatory therapeutics in the context of chronic infection as standardized by the Cystic Fibrosis Foundation. The reason the model is versatile is that it aids in determining whether anti‐inflammatory therapeutics affect infection exacerbation. , , In this article, the model is used to define hMSC anti‐inflammatory and antimicrobial potency, which could be translated to other diseases besides CF that are associated with NTM infections.

Bone marrow‐derived macrophage culture

Bone marrow‐derived macrophages (BMDMs) were derived from WT and CF mice as previously described. , , Briefly, hematopoietic cells from the bone marrow were rinsed with 1× phosphate buffered saline and centrifuged at 1800 rpm for 9 minutes. After the supernatant was discarded, cells were resuspended with medium (Roswell Park Memorial Institute medium (RPMI) + 10% heat inactivated fetal bovine serum (HI‐FBS) + 1% penicillin‐streptomycin (PSG) + 46mL L929 generated conditioned medium) and plated at 2.5 × 106 cells in 5 mL per petri dish.

NTM culture conditions

M. avium (#49601; American Type Culture Collection, Manassas, Virginia) and/or M. intracellulare (#13950; American Type Culture Collection) were suspended using 7H9 broth. From the resuspended bacteria, 1 mL was used to grow bacterial culture in a 125‐mL Erlenmeyer flask with 12 mL broth (7H9 broth, #M0178‐500 g; albumin‐dextrose‐catalase (ADC) enrichment, #M0553‐1 VL; Sigma, St. Louis, Missouri) and was incubated at 37°C with 5% CO2. Culture medium was changed every 5 days with 10 mL fresh 7H9 broth to continue growth. Optical density was monitored at every collection and medium change to ensure proper pathogenic concentration. M. avium complex and the participating NTM pathogens are highly variable depending on the isolate. M. avium and M. intracellulare are generally the dominant species. The model used to mimic Mycobacterium avium complex (MAC) consisted of 50% M. avium and 10% M. intracellulare.

In vivo models of NTMs

M. avium and M. intracellulare were cultured 1 week by adding 1 mL of bacterial stock to 12 mL of Middlebrook 7H9 broth as extensively described previously. , One week later, bacterial optical density was measured (600 nm) to define the concentration of either NTM pathogen to put directly into mice (acute infection/inflammation model) or into the bead preparation (sustained infection model) defined by the optical density/colony forming unit (CFU) relation of either M. avium or M. intracellulare. All mice had spleen, liver, and lung CFU quantification.

Acute infection

Mice were anesthetized with isoflurane followed by instilling, through the intranasal passageway, 100 μL of 106 CFUs of NTMs.

Sustained infection

NTMs were embedded into agarose beads as previously described. The NTM bead preparations were diluted 1:50, 1:500, 1:5000, and 1:50 000 and plated on 7H9 agar plates (M9029; Teknova, Hollister, California ) to validate the dosing/bead preparation prior to use in vivo. Agarose beads were prepared and verified for CFUs to verify NTM dosing. , , Once the CFUs for the pathogen bead preparations were known, beads were diluted so that each animal received 105 CFUs in 50 μL using transtracheal infusion. After surgery, the diluted solution was plated once more on 7H9 plates and incubated overnight to validate infection with dosing the mice. Mice were monitored each day up to day 7 by tracking individual weights and thermal imaging for clinical score. Serum and bronchoalveolar fluid were obtained at termination for monitoring inflammation and infection status. Mouse models are notoriously excellent at resolving pulmonary infections, which is why the agarose bead model is used to establish a long‐term infection. , The beads themselves degrade over time, which is essential because foreign bodies in the lung create granulomas and foreign body reactions. , Free bacteria are resolved within 24 to 48 hours, so the use of the agarose bead model allows for extension of the infection and transition of the inflammatory response. In this article, the 7‐day NTM model is used to establish hMSC potency in vivo.

Thermal imaging

Individual mice were placed in a polyvinyl chloride (PVC) ring roughly 4.5 inches tall, on a nonreflective surface kept at constant temperature. An infrared camera (FLIR, Wilsonville, Oregon) was held above the ring at a standard height, and a picture was taken. Photos were then analyzed by converting to a series of temperature measurements at specified locations on the body of the mouse using FLIR Tools software. Measuring a series of external temperatures of the mouse allows for an estimate of internal temperature, which in turn provides a clinical score to describe overall inflammation and overall health. The clinical score ranges from 0 to 12, with a higher score suggesting better survivability and lower overall inflammation. A score of 4 or below is indicative of poor infection resolution and a decreased probability of next‐day survivability. These clinical scores are comparable to other observable metrics of health such as weight and postmortem bacterial load. The thermal imaging was used to evaluate the acute response to infection out to 3 days and the sustained response out to 7 days as verified by CFUs and weight changes.

Human mesenchymal stem cells

Human posterior iliac crest bone marrow samples were obtained after written informed consent with Case Western Reserve University and University Hospitals Internal Review Board (#CASE12Z05). hMSCs were isolated and expanded in ex vivo culture per previously published methods. , , Pre‐good manufacturing practices (GMP) procedures have also been previously described. hMSCs (passage 2 or 3) were grown in antibiotic‐free conditions for 3 days prior to harvesting the conditioned medium (supernatant) or cells for use in these studies. Table 1 details the deidentified hMSC preparations for age range, demographics, and gender that were made available. hMSC antibiotic‐enhancing activity involved culturing the hMSCs and gentamicin (10 ng/mL) concurrently with M. avium, M. intracellulare, or MAC CFUs.

TABLE 1

hMSCs donor sources profiled for anti‐NTM studies

Donor	Gender	Age (years)
A	Female	40‐50
B	Male	30‐40
C	Female	20‐30
D	Not provided	Not provided
E	Not provided	Not provided
F	Male	20‐30
G	Not provided	Not provided
H	Not provided	Not provided
I	Not provided	Not provided
J	Female	30‐40
K	Male	30‐40
L	Female	40‐50
M	Male	30‐40
N	Male	30‐40
O	Male	40‐50
P	Male	20‐30
Q	Male	20‐20
R	Male	30‐40

Luminex assay

Assays were obtained from R&D Systems (Minneapolis, Minnesota) and done according to the manufacturer's instructions using a Luminex® 100/200™ System 0/200 evaluating the following: tumor necrosis factor (TNF)‐α, IL‐6, IL‐1β, and keratinocytes‐derived chemokine (KC). BMDMs (n = 3 different preparations) in duplicate were treated with M. avium or M. intracellulare either as a free solution or embedded into agarose beads. Cell culture supernatants from each of the BMDM experiments were harvested 24 hours later, aliquoted, and frozen prior to the Luminex assays. The multiplex assay was processed according to the manufacturer's directions using a Five Parameter Logistic (5PL) curve fit. , LL‐37 was quantified using an enzyme‐linked immunosorbent assay as previously described using our Clinical and Translational Science Collaborative (CTSC) Bioanalyte Core, which is directed by Dr Bonfield. ,

Reverse transcription polymerase chain reaction

mRNA extraction was performed through phenol extraction followed by cDNA to monitor gene expression using reverse transcription polymerase chain reaction. Quality of mRNA and cDNA was assessed through NanoDrop spectrophotometry (optimal threshold 260‐280 nm; Thermo Fisher Scientific, Waltham, Massachusetts). , , Validation of cDNA quality was done through use of a reference gene peptidyl prolyl isomerase (hPPIA; Hs04194521_s1). Validated primers were obtained from Applied Biosystems (Foster City, California) for IL‐6 (Hs00174131_m1), HMOX‐1 (Hs01110250_m1), and CCL20 (Hs00355476_m1), which were normalized to the validated internal expression of hPPIA. Data is expressed as fold change of IL‐6, HMOX‐1, CCL20 gene threshold cycle value as compared to the housekeeping hPPIA gene.

Statistical analysis

Analyses of linear or log transformation were used to compare between experimental conditions using unpaired t tests, and one‐way analysis of variance (ANOVA) quantified assays also used comparisons of slopes over time. , In the acute and chronic infection models, survival curves were compared using stratified log‐rank tests with pathogens M. avium and M. intracellulare as strata. When comparing cytokine levels, paired t tests were used, assuming that SDs of 0.75 on the log2 scale by a 2.8‐fold differences in expression can be detected with 80% power with two‐sided tests at the .01 significance level. Pathology (eg, bacterial load, white blood cell counts, and cytokines) was log‐transformed as necessary to compare between groups or conditions using one‐ or two‐way ANOVA, treating donors as experimental blocks. Prism software (version 9.0.2; GraphPad Software, La Jolla, California) was used for analyses. One‐way ANOVA and Bartlett's correction were incorporated to manage multiple variables and correction for data normalization when required. All significance was defined by the 95% confidence interval at P ≤ .05.

RESULTS

hMSC anti‐NTM potency

We have previously described a systematic way to follow the growth, metabolism, and survival of M. avium and M. intracellulare. We have also described the potential for hMSCs and their secreted products to have antimicrobial and anti‐inflammatory potential. , , Given these findings, we focused on the potential of hMSCs for anti‐nontuberculous mycobacterial activity. To begin these studies, we first performed in vitro assays to determine whether hMSCs and their secreted products had anti‐nontuberculous mycobacterial potential and whether the supernatants and cells had differential effects on the nontuberculous mycobacteria (Figure 1). We found the hMSCs and their supernatants to have differences in their ability to affect MAC (Figure 1A), M. intracellulare (Figure 1B), and M. avium (Figure 1C) CFUs in vitro over 24 hours. hMSCs significantly decreased CFUs in MAC, trended toward significance in M. intracellulare, and significantly decreased M. avium CFUs. Supernatants decreased CFUs in all conditions but were only significantly effective in MAC (Figure 1A). Given these findings, we wanted to evaluate the effect of the hMSCs on the nontuberculous mycobacteria over a longer time span as well as to assess the impact of different hMSC donor preparations.

FIGURE 1

hMSC cell and supernatant anti‐nontuberculous mycobacterial potency. hMSC growth medium, hMSC cells, or hMSC supernatant (antibiotic‐free 72‐hour conditioned medium) was cultured with MAC (A), M. intracellulare (B), or M. avium (C) for 24 hours. Conditions were plated on 7H9 agar plates and evaluated 3 days later. hMSC cells and supernatants significantly decreased MAC CFUs (A; P < .005 and P < .0005, respectively) and CFUs compared with hMSC growth medium (P < .05 and P < .005, respectively). There was also a significant difference between hMSC cells and supernatants in the effects on MAC CFUs (P < .05). The hMSCs and their supernatants did not have a statistically significant effect on M. intracellulare CFUs (B); however, hMSCs trended toward significance (P = .0585). In M. avium, the hMSCs were the only group to have a statistically significant effect on CFUs (C; P < .05). CFU, colony forming unit; hMSC, human mesenchymal stem cell; MA, M. avium; MAC, M. avium complex; MI, M. intracellulare; sup, supernatant. ns = P > .05, *P< .05, **P< .005, ***P< .0005

Supernatants from 12 different donor preparations of hMSCs were cultured with MAC, M. intracellulare, and M. avium. Figure 2A focuses on how each of the hMSC supernatant preparations affects the survival of MAC from 24, 48, and 72 hours of treatment as defined by CFUs compared with MAC without hMSC treatment (culture medium control). This panel demonstrates the percent decreased number of CFUs of each preparation of MAC without and with hMSC supernatant treatment. Each color represents a different hMSC donor preparation. The ability to decrease MAC growth was variable by hMSC preparation, with some hMSC supernatants approaching 100% capacity to kill MAC and others showing much lower capacity to kill MAC, across 24, 48, and 72 hours of treatment. With culture conditions were extended to 72 hours, we could evaluate the hMSC preparations for sustained anti‐MAC activity, which showed that the sustainability of the anti‐NTM effect was also hMSC donor dependent (Figure 2A). These data demonstrate that each of the hMSC donor supernatants exhibits different antimicrobial potency and potential for sustainable efficacy in vitro.

FIGURE 2

hMSC supernatant anti‐nontuberculous mycobacterial long‐term potency. A, hMSC supernatants from antibiotic‐free 72‐hour conditioned medium were co‐cultured with MAC. Cultures were allowed to proceed for 3 days with a daily evaluation of pathogen burden by CFUs. Bacteria were plated on 7H9 plates and incubated for 3 days to allow for evaluation of the colonies. The majority of the hMSC preparations had anti‐MAC activity, but the kinetics of the anti‐MAC activity changed over time and were donor dependent. B, hMSC supernatants from antibiotic‐free 72‐hour conditioned medium were co‐cultured with M. intracellulare. Cultures were allowed to proceed for 3 days with a daily evaluation of pathogen burden by CFUs. The majority of the hMSC preparations had anti‐MI activity, but the kinetics of the anti‐MI activity changed over time, and there was large donor variability. C, hMSC supernatants from antibiotic‐free 72‐hour conditioned medium were co‐cultured with M. avium. Cultures were allowed to proceed for 3 days with a daily evaluation of pathogen burden by CFUs. The majority of the hMSC preparations had anti‐M. avium activity with the anti‐M. avium activity changing with time and hMSC donor preparation. CFU, colony forming unit; hMSC, human mesenchymal stem cell; MAC, M. avium complex

Because MAC is composed of both M. avium and M. intracellulare, the next series of studies explored how donor hMSC preparations affected each NTM individually, hypothesizing that the variability may be related to hMSC donor‐specific NTM‐specific antimicrobial activity. To begin to determine if the variability observed in the hMSC anti‐MAC potency was due to a specific species of nontuberculous mycobacteria, each of the hMSC preparations was screened against M. intracellulare (Figure 2B) and M. avium (Figure 2C) separately. The results demonstrated that there was significant variability in the ability of the hMSC supernatants to affect M. intracellulare or M. avium CFUs. Additionally, the kinetics of the different hMSCs donor preparations on M. intracellulare (Figure 2B) and M. avium (Figure 2C) were substantially variable. The hMSC supernatant donors each had their own profile, suggesting that selection of the hMSC preparation for treatment would be based upon which pathogen was being treated and whether the application required an immediate or sustained antimicrobial effect. The differences between hMSC supernatant effects on M. avium and M. intracellulare would likely define anti‐MAC potency because M. avium is the dominant species in MAC. If a preparation is not efficient against M. avium but kills M. intracellulare, it may not have the same clinical benefit as an hMSC preparation that is potent against both if a patient is infected with MAC, rather than a singular NTM.

Antimicrobial potency and therapeutic efficacy are also defined by the ability to enhance sensitivity to antibiotics. In previous studies, we have shown that hMSCs and their secreted products have the capacity to enhance antibiotic effects. To determine if the hMSC supernatants would affect the sensitivity of the NTMs to antibiotics, we cultured the hMSCs with MAC, M. avium, or M. intracellulare with and without the addition of gentamicin. Although gentamicin is not a macrolide, it does have anti‐NTM potency and provides a good indication for the potential for antibiotic‐enhancing potential based upon our previous work. , Four of the preparations highlighted in Figure 2 were tested in the presence and absence of suboptimal concentrations of gentamicin (10 ng/mL; Figure 3). There was significant variability in the antibiotic‐enhancing potency of each hMSC donor preparation, with two of the hMSC preparations having high antibiotic‐enhancing potential approaching 70% to 80% at 72 hours for MAC (Figure 3A). The other two preparations provided action against MAC activity, but the potency was not sustainable. Evaluating the same hMSCs against the individual pathogens M. intracellulare and M. avium demonstrated the overall capacity to enhance sensitivity to antibiotics. In M. intracellulare, there was sustained impact of the hMSC supernatant and gentamicin on CFUs, with 40% to 60% maximum at 72 hours (Figure 3A). The same two preparations that had the highest potency against MAC and M. intracellulare (Figure 3; hMSC1 and hMSC2) also had the highest potency against M. avium (Figure 3C) approaching 60% to 80% at 72 hours. The antibiotic‐enhancing capacity of the two best preparations was much higher in MAC and M. avium than in M. intracellulare, consistent with the dominance of M. avium in MAC.

FIGURE 3

hMSC nontuberculous mycobacterial antibiotic‐enhancing potency. hMSC supernatants from antibiotic‐free 72‐hour conditioned medium were co‐cultured with MAC (A), M. intracellulare (B), or M. avium (C) with and without concurrent addition of 50 μg/mL gentamicin. Cultures were allowed to proceed for 3 days with a daily evaluation of pathogen burden by CFUs. Bacteria were plated on agar plates and incubated for 3 days to allow for evaluation of the colonies. The majority of the hMSC preparations had antibiotic‐enhancing potency, but the impact was variable depending on the pathogens evaluated. There was large variability in effectiveness over time between donors. CFU, colony forming unit; hMSC, human mesenchymal stem cell; MAC, M. avium complex

In vivo models of NTM infection

To transition in vitro studies to in vivo models, we used the WT and CF pathogen modeling systems to develop a standardized in vivo model of NTM infection. We chose to examine M. intracellulare and M. avium separately for these studies to better understand the impacts of each pathogen before proceeding into dual pathogen in vivo studies. Initially, a 106 CFU inoculum of M. avium or M. intracellulare was administered intranasally into WT and CF mice (Figure 4). WT mice resolved the M. intracellulare and M. avium infection better than the CF mice, based on CFU bacterial load in the lung (Figure 4A,B). None of the mice had spleen or liver CFUs per our assay of these tissues. However, both WT and CF mice lost weight with the M. intracellulare and M. avium infection but resolved weight loss at similar rates by day 3, consistent with resolving an infection (Figure 4C,D). To enhance our ability to monitor the infection without euthanatizing the mice, we used our thermal imaging technology to monitor the infection response longitudinally in the mice. The thermal imaging technology provides insight into the host response in vivo. The data that are obtained from the thermal image transform the thermal heat map into a quantitative clinical score with minimal stress to the mice. The imaging profiles of a WT and CF mouse infected with M. intracellulare (Figure 4E) or M. avium (Figure 4F) are shown. Day 0 is the thermal image of the mice prior to the free M. intracellulare or M. avium infection along with their weight and thermal score. After infection, the mice were imaged on days 1, 2, and 3 prior to euthanasia. With M. intracellulare, the representative WT mouse loses minimal weight on day 1 and regains weight and improves thermal score by day (Figure 4E). The representative CF mouse loses weight and has a decreased thermal score until day 3 (Figure 4E). With M. avium, the representative WT mouse stays consistent in both weight and thermal score, whereas the representative CF mouse stays consistent in weight and thermal score early on but loses weight and decreases thermal score on the last 2 days (Figure 4F). These changes in thermal score and weight with M. intracellulare and M. avium infection were profiled and compared across the experiments to demonstrate the accuracy of the thermal imaging technology in predicting inflammation status and clinical score (Figure 4G,H). The studies emphasize both our efficiency at infecting the mice with the different NTM species in an acute setting of infection and the fact that in the acute setting there is little, albeit some, difference in the infection trajectory between WT and CF mice infected with NTMs. NTM infections do not tend to be acute but consistently transition to more insidious and long‐lasting infections in CF.

FIGURE 4

Free M. avium and M. intracellulare in vivo model. WT and CF mice were infected with 106 CFUs of free M. intracellulare (A), M. avium (B), and followed for 3 days for CFUs, weight, and thermal score. A,B, Mean CFUs 3 days after infection for M. intracellulare and M. avium, respectively, in WT and CF mice. C,D, Percent weight change over the course of the study compared with baseline weight prior to infection for M. intracellulare (C) and M. avium (D). E‐H, Thermal imaging of the mice on each day of the experiment was performed using FLIR camera. Representative thermal images with corresponding weight and thermal score are shown for M. intracellulare (E) and M. avium (F). Thermal image scores were compared with percent weight change over the course of the study to demonstrate accuracy of thermography scoring system for M. intracellulare (G) and M. avium (H). CF, cystic fibrosis; CFU, colony forming unit; MA, M. avium; MI, M. intracellulare; WT, wild type. ∗P < .05

hMSC therapeutic development to treat chronic NTM infections requires a model that can recapitulate a longer infection and inability to effectively resolve the NTM infection, such as that is seen in human disease. To address this need, we adopted our standardized agarose bead infection modeling system, which we use to evaluate antipathogen and anti‐inflammatory therapeutics for potency and efficacy to NTM infection. , , For these studies, we embedded M. intracellulare and M. avium with a known viable inoculum of 106 CFU of bacteria (based upon optical density and CFUs) into an agarose/bead‐mineral oil matrix of 5 to 10 μm. Pathogens can respond to the modeling matrix differently in terms of sensitivity to heat and manipulation thus verifying that the process of preparing the agarose beads did not alter the growth of the NTMs by culturing the agarose bead preparations embedded with M. intracellulare or M. avium to evaluate the survival and pathogenic load of each species (Figure 5A). After assuring the capacity of the pathogens to survive in the modeling technology, we translated the bead count of CFUs/bead numbers to consistently deliver 106 CFUs for the in vivo mouse model. The next goal was to determine if the M. intracellulare and M. avium bead preparations could induce a sustained infection in vivo using the CF and WT mice. Using the validated agarose bead‐NTMs preparations in Figure 5A, we infected WT and CF mice with M. intracellulare or M. avium and followed the mice for 7 days. Both WT and CF had sustained infection at day 7 (Figure 5B,C). The studies demonstrate the ability to generate a more sustained in vivo infection model of M. intracellulare and M. avium and highlight the differences in the infection trajectory between the WT and the CF mice.

FIGURE 5

Development of the in vivo sustained M. intracellulare and M. avium infection model. A, Using growth conditions optimized for M. intracellulare and M. avium, we validated the ability to embed into agarose beads. These beads were tested for potency; results are shown in Table 2. B,C, The same bead preparations were used to infect C57BL/6J or Cftr tm1Kth knockout mice. Mice were euthanized, and whole lung homogenate lung CFUs were defined for M. intracellulare (n = 20) (B) and M. avium (n = 10) (C). D,E, Wild‐type (WT) and cystic fibrosis model (CF) mice were infected with M. intracellulare (D) or M. avium (E) beads, and a group of CF mice was also given hMSC cells retro‐orbitally to determine potency of hMSCs on nontuberculous mycobacteria (NTM) infection in vivo, as defined by lung homogenate CFUs. hMSCs significantly decreased M. intracellulare CFUs in the CF mice (D; P < .05). In the M. avium infection model, hMSC significantly decreased CFUs in the CF mice (E; P < .0001) and even decreased the CFUs of the CF mice compared with WT (P ≤ .0001). None of the mice had detectable spleen or blood NTM CFUs. *P< .05, ****P< .0001, CFMA, CF mice infected with M. avium treated; CFMI, CF mice infected with M. intracellulare; CFMA + hMSC, CF mice infected with M. avium treated with hMSCs; CFMI, CF mice infected with M. intracelluare treated with hMSCs; CFU, colony forming unit; hMSC, human mesenchymal stem cell; WTMA, wild type mice infected with M. avium; WTMI, wild type mice infected with M. intracellulare

TABLE 2

Bone marrow‐derived macrophage response to nontuberculous M. avium and M. intracellulare

Cytokine	Unstimulated		MA free		MI free		Sterile beads		MA beads		MI beads
	Cftr+/+	Cftr−/−	Cftr+/+	Cftr−/−	Cftr+/+	Cftr−/−	Cftr+/+	Cftr−/−	Cftr+/+	Cftr−/−	Cftr+/+	Cftr−/−
IL‐6	5 ± 2	3.5 ± 0.5	115 ± 1.5	137 ± 3	432 ± 12	591 ± 11	37 ± 2.5	220 ± 58	2372 ± 188	7460 ± 488	43 ± 122	2365 ± 214
TNFα	1 ± 0.1	1 ± 0.01	379 ± 30	461 ± 22	641 ± 128	836 ± 70	371 ± 199	1412 ± 636	746 ± 58	888 ± 185	74 ± 372	836 ± 70
IL‐1b	11 ± 4	15 ± 1	109 ± 5	137 ± 3	229 ± 20	591 ± 11	74 ± 23	220 ± 58	3480 ± 648	7460 ± 488	914 ± 274	2365 ± 214
KC	120 ± 5	103 ± 2	3383 ± 14	2836 ± 208	9935 ± 152	7911 ± 255	240 ± 40	900 ± 32	4280 ± 148	2200 ± 140	3294 ± 27	2804 ± 17

Note: P < .05; n = 3. All values are pg/mL, means ± SEM. Samples run in duplicate using Luminex Multiplex with 50 beads per cut; values were quantified after subtraction of the blank and verified for minimal nonspecific reactivity in multiplex assay.

Abbreviations: Cftr, cystic fibrosis transmembrane conductance regulator; Cftr+/+, Cftr wild type; Cftr−/−, Cftr knockout; IL, interleukin; KC, keratinocytes‐derived chemokine; MA, M. avium; MI, M. intracellulare; TNFα, tumor necrosis factor α.

Therapeutic efficacy of the different donor hMSCs in vivo

With the establishment of the model, we wanted to test if the donor hMSC potency differences in vitro correlated to in vivo efficacy. Both hMSC cells and the supernatants have anti‐NTM activity (Figure 1). hMSCs manage the surrounding milieu through the production of soluble mediators, which is why each of the hMSC donor supernatant preparations was screened against the NTMs (Figure 1). However, the in vivo models receive intact cells and not the supernatants. Based on the results of the beginning studies, the most efficacious hMSC donor (106/100 μL given through the retro‐orbital sinus) was used to treat WT and CF mice infected 24 hours earlier with agarose beads embedded with 105 CFUs of either M. intracellulare (Figure 5D) or M. avium (Figure 5E). In evaluating the in vivo bead models, the CF mice consistently had higher CFUs relative to the WT mice at day 7 (Figure 5D,E) regardless of the pathogen. None of the mice suffered from NTM decompartmentalization, quantified by negative spleen, liver, and blood CFUs. The administration of hMSCs in the CF mice resulted in significantly decreased lung CFUs of both M. intracellulare and M. avium infection compared with CF mice not treated with hMSCs at day 7 (Figure 5D,E). Again, mice had negative spleen, liver, and blood CFUs. In the M. intracellulare bead model, all mice had sustained inflammation in response to infection; however, the CF mice had significantly higher lung CFUs at 7 days compared with WT or CF treated with hMSCs (Figure 6A). All M. intracellulare bead‐infected mice lost weight over the course of the 7 days; however, the hMSC‐treated CF mice brought the percent weight change back to that of the WT (Figure 6B). Thermal imaging was performed on the NTM bead models over the course of the 7 days, starting from the day prior to infection. Figure 6C demonstrates the thermal imaging of representative M. intracellulare bead‐infected WT, CF, and CF + hMSC mice with correlating thermal score and weight. The representative WT mouse lost weight on day 4 postinfection but regained weight by day 7, which correlated with thermal score (Figure 6D). The representative CF mouse lost significant weight over all 7 days of the infection study and had a similarly decreasing thermal score commensurate with weight loss (Figure 6E). The representative CF mouse given hMSCs lost weight during the first 4 days postinfection and started to regain weight on day 7, which correlated with thermal score on days 0 through 7 (Figure 6F). These studies indicate that hMSCs have the capacity to attenuate the NTM infection course in the CF mice, attenuating lost weight, thermal score, and pathogenic load. MAC was not pursued in the agarose bead model at this time because the ratio of M. intracellulare and M. avium is variable between different MAC isolates and the relative sensitivities of M. intracellulare and M. avium to the different hMSC preparations were different. Future studies are aimed at combining agarose bead preparations so that changes in exact M. intracellulare and M. avium ratios can be incorporated.

FIGURE 6

In vivo sustained M. intracellulare infection model treated with hMSCs. WT and CF mice were infected with M. intracellulare beads, and a subset of CF mice was given hMSC cells retro‐orbitally. Mice were followed for 7 days for weights and thermal imaging, and postmortem CFUs were collected. A, CFUs from the WT, CF, and CF + hMSC mice were determined from whole lung homogenate. hMSCs significantly decreased M. intracellulare CFUs in the CF mice, but there was wide variability in this reduction (D; P < .05). B, Weight change was evaluated as a percent change from baseline weight prior to infection. C, Thermal imaging was performed on all mice starting the day before infection. Thermal images of representative WT (top), CF (middle), and CF + hMSC (bottom) mice, along with corresponding weights and thermal scores are shown. D‐F, Thermal image scores were compared with percent weight change (for representative mouse) over the course of the study to demonstrate accuracy of thermography scoring system. None of the mice had detectable spleen or blood nontuberculous mycobacterial CFUs. *P < .05, CF, cystic fibrosis; CFMI, CF mice infected with M. intracellulare; CFU, colony forming unit; CFMI+hMSCs = CF mice infected with M. intracellulare treated with hMSCs; hMSC, human mesenchymal stem cell; MI, M. intracellulare; WT, wild type; WTMI, wild type mice infected with M. intracellulare; WTMI+hMSCs, wild type infected with M. intracellulare treated with hMSCs

Nontuberculous mycobacterial bead preparations and macrophage response

The goal of the agarose bead preparation was to sustain the pathogen in vivo and to provide a potent NTM signal for an immune response. Nontuberculous mycobacteria are intracellular pathogens that infect macrophages as the primary site of intrusion. To verify that the process of embedding the NTM into the agarose beads did not alter the signal potency of the pathogens, the agarose bead‐NTM preparations were cultured with BMDMs, with the BMDMs analyzed for the inflammatory response. CF BMDMs are known to have a more robust response than WT BMDMs and have consistently been used to follow changes in immune response in vitro. , , To evaluate the potency of the agarose bead preparations to induce BMDM cytokines, we compared the proinflammatory potency of the agarose beads embedded with M. avium or M. intracellulare with that of free M. avium or M. intracellulare, with lipopolysacchride (10ug/mL) and sterile agarose beads as positive and negative controls respectively. Table 2 shows the BMDM response to free bacteria and agarose beads embedded with bacteria.

Overall, there were pathogen‐specific and cytokine‐specific responses to each of the different agarose bead‐NTM or free NTM conditions. The free M. avium and M. intracellulare induced a higher proinflammatory chemokine level (KC) in the CF BMDMs then the M. avium and M. intracellulare beads, which is likely due to the acute nature of the free infection. The NTM‐laden beads induced higher IL‐1β expression in the CF BMDMs than in the WT, which is most likely due to the more chronic nature of the infection inducing macrophage activation. IL‐6 expression was higher with the NTM‐laden beads with CF BMDMs, which is consistent with CF pathophysiology and hyperproduction of IL‐6 in chronic disease. TNF was elevated but not different between WT or CF in response either acute or chronic infection. In all cases, the CF BMDMs had higher levels of secreted proinflammatory cytokines in response to the NTM bead preparations than WT BMDMs (in Table 2; P < .05), consistent with the CF response.

Transcriptional analysis for genes defining hMSC potency in response to NTM

We wanted to use hMSC response as a means to correlate hMSC potency and efficacy to begin to understand what might differentiate a specific hMSC preparation for potency in treating NTMs. As a follow‐up to previous studies, LL‐37 was quantified in the supernatant derived from the hMSCs, showing significant variability but no association with anti‐NTM potency (mean ± SD, 17192 ± 9345 pg/mL; n = 12; range, 5344‐32 379 pg/mL). The hMSCs that generated the supernatants screened in the antimicrobial assays presented in Figures 2 and 3 were evaluated for other biological response mediators suggested to contribute to the management of NTM infections: IL‐6, HMOX‐1, and CCL20 (Figure 7A‐C), using gene expression profiling. TNFα was not pursued because the cytokine levels were not different between the WT and CF BMDM response to pathogens (Table 2). In these experiments, each of the different hMSC preparations was treated with either free M. intracellulare or M. intracellulare embedded onto agarose beads and followed for their gene expression profiles. When each hMSC preparation was evaluated individually (Figure 7), there was significant differences in overall response whether the M. intracellulare was free or embedded in the agarose beads (P ≤ .05 comparing hMSC with pathogen vs hMSC without pathogen for each preparation individually). There was no detectable LL‐37 transcriptional response (data not shown). These studies suggest that the hMSC cytokine gene expression signature may be able to define anti‐NTM potency, if gene expression profiles predict therapeutic functionality.

FIGURE 7

hMSC biomolecule inflammation and nontuberculous mycobacteria (NTM) infection modifiers. hMSCs from the donors profiled for anti‐nontuberculous mycobacterial potency were cultured directly with M. intracellulare in a free slurry or embedded in agarose beads. Each of the hMSC donors was evaluated for cytokine gene expression post‐mRNA/cDNA processing, focusing on biomolecules implicated in NTM management. hMSCs were screened for three main genes, IL‐6 (A), hHMOX (B), and CCL20 (C). Each different preparation had a unique gene expression fingerprint for these specific proteins. The mean values were not different between the different treatment groups for each of the genes given the diversity in the hMSC fingerprint. Each individual hMSC preparation was treated with either free MI or MI embedded in agarose beads but served as its own control; the hMSC donor phenotype becomes apparent. The treatment profiles for each donor were statistically significant for individual donor responses in most cases, especially with regards to IL‐6 (P <.05) and CCL20 (P < .05). ΔCT, cycle threshold; CCL20, macrophage inflammatory protein 3α; hHMOX, human heme oxygenase; hMSC, human mesenchymal stem cell; IL, interleukin; MI, M. intracellulare

DISCUSSION

NTM infections can be chronic and very difficult to resolve. Based upon our previous work, we have delineated the anti‐inflammatory‐, antimicrobial‐, and antibiotic‐enhancing potential of hMSCs to aid in treating chronic nontuberculous mycobacterial infections. , , In our initial studies, we focused on defining the anti‐NTM potency of hMSCs as well as the variability between different hMSC preparations. The results of our findings demonstrated the significant variability and potency of hMSCs against MAC (combined M. intracellulare and M. avium) as well as M. intracellulare and M. avium individually. These studies ultimately point out that not all hMSC preparations have the same level of antimicrobial potency or sustainability in activity, suggesting that it is essential to identify the appropriate hMSC donor and subsequent preparation for disease‐specific applications.

To refine these criteria, we developed in vitro and in vivo models of M. avium and M. intracellulare in which we tested hMSC preparations for anti‐NTM activity. In the validation of the hMSCs, we used the CF pulmonary phenotype to define disease in response to NTM infections. The CF phenotype has an increased susceptibility to the NTM infections as well as enhanced inflammatory response to pathogens. We further standardized in vivo models of both acute and sustained M. intracellulare and M. avium infection. The acute model did not generate a long enough infection to evaluate the hMSC potency. Using our extensive experience with developing in vivo modeling, agarose beads were used to sustain NTMs in the lungs to provide an in vivo setting for testing hMSC anti‐NTM potency. The studies to directly correlate different hMSCs in both the WT and the CF models are ongoing with the ultimate goal of defining the functional variability in vitro with in vivo endpoints. These studies, although they are in the early stages, are the first of their kind against NTMS.

Up to 7% of Western society suffers from a chronic inflammatory disease. Rates of inflammatory conditions, including diabetes, autoimmune disorders, cardiovascular diseases, and respiratory diseases, have increased significantly in the last three decades. Chronic inflammation is often concurrent with infection, increasing mortality rates among suffering individuals. , NTMs colonize individuals with chronic lung diseases, and the treatment options for patients suffering from these diseases include immunotherapies, steroids, macrolides, and heavy antibiotic courses. Many of these treatments are ineffective or have significant and long‐lasting side effects such as antibiotic resistance, heart enlargement, or the development of cancer. , , Patients who suffer from chronic and insidious infections with NTMs must deal not only with the consequences of the disease but also the toxicity and side effects of the long‐term therapeutic intervention. , The translation of a new therapeutic from laboratory innovation to a clinical setting requires a stepwise development and preclinical validation before use in patients. Models that mimic what happens in sustained NTM infections have been difficult to establish because small animal models clear bacteria quickly and larger animals are too expensive for defining dosing, timing, and duration of a new treatment. , The numbers required to power these types of studies make them costly. This article demonstrates the development of an in vivo mouse model of sustained NTM infection, using a standardized agarose bead model and a variety of different technological approaches to validation. , Using this modeling system, this article demonstrates a means by which NTM infections can be sustained in vivo. Future studies are ongoing to define a long‐term model for therapeutic testing.

hMSCs are a relatively new and exciting therapeutic option for the treatment of chronic inflammation and infection given their unique capacity to respond and actively contribute to their host environment. The differences between hMSCs and other pharmaceutical therapeutics lies in the “dynamic” nature of hMSC paracrine repertoires responsible for clinical effectiveness. Many subjects enrolled in clinical trials involving hMSC therapeutics experience dramatic clinical success; however, there is a significant nonresponder population that does not obtain benefits from the same therapy. The effectiveness of hMSC therapeutic contributions is dictated by disease severity and the hMSC phenotype contributing to an inefficient hMSC therapeutic. The validation of “bench‐to‐bedside” modeling systems to mimic chronic NTM is essential to identify the best culturing conditions that result in optimal and reproducible assays for hMSC clinical potency for the disease‐specific pathophysiology for treating diseases such as NTM lung infections.

hMSCs have the potential of managing infection and inflammation through both direct and indirect interactions with the host. The production of soluble antibacterial mediators like LL‐37 and CCL20 directly support immune cell recruitment and antimicrobial activity. , , , These antimicrobial peptides can define the phenotype of the initial host response to infection as well as directly interact with the pathogen, increasing susceptibility to antibiotics and other pharmaceuticals aimed at slowing or killing the bacteria directly. Our studies here did not find correlation between the production of LL‐37 and anti‐NTM activity. LL‐37 is highly complex with the capacity to drill pores into extracellular pathogens and provide a chemotactic gradient for cellular recruitment. , NTMs are intracellular pathogens, using macrophages to optimize pathogenicity and potentially protecting themselves from host antimicrobial activity. , LL‐37 may be very important in providing efficiency in immune clearance of NTMs but may not have a direct impact on pathogen survival. , This represents collaborative work ongoing in our laboratory. To follow up on these studies, we focused on monitoring soluble factors that have been directly linked to NTM management in the literature. , hMSC mediators, such as HMOX‐1, indoleamine‐pyrrole 2,3‐dioxygenase (IDO), IL‐17, IL‐6, IL‐8, and IL‐10, can redirect host immunity by affecting macrophage and T‐cell phenotypes for specific adaptive and innate mechanism to support the host in the pursuit of homeostasis. , , , Depending on the host milieu and the unique phenotype of the hMSC donor, it is easy to understand the variability in host responsiveness to cell‐based hMSC therapy. Every donor hMSC preparation has a unique profile in terms of response to pathogens, which likely translates into the downstream events related to successful potency and host response to treatment. Future studies are planned to evaluate unique hMSC fingerprints associated with antimicrobial potency and whether it is global or pathogen specific.

CONCLUSION

Sustained infection and inflammation are often associated with infection with bacteria such as NTM species M. avium and M. intracellulare. hMSCs are therapeutic tools that have been studied in more than 1200 clinical trials associated with some aspect of inflammation and infection (ClinicalTrials.gov). hMSCs are unique in their capacity to respond by secreting multiple bioactive factors and actively contributing to the host environment; this gives the hMSCs a clinical advantage over traditional pharmaceuticals. , In this article, a strategy has been developed focusing on selecting an hMSC product that had the highest in vitro and in vivo preclinical potency against M. intracellulare and M. avium. Furthermore, through effector modification of hMSCs and clinically translational modeling systems, the studies in this article will be able to bridge the gap between bench validation and clinical implementation of the “ideal” hMSC donor and resultant phenotype for the treatment of pulmonary diseases complicated with NTM mycobacterial infections.

CONFLICT OF INTEREST

The authors declared no potential conflicts of interest.

author contributions

T.L.B.: conception/design, financial support, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript; M.S., M.F.: collection and/or assembly of data, data analysis and interpretation, manuscript writing; D.F.: collection and/or assembly of data, data analysis and interpretation; V.R.: collection and/or assembly of data, manuscript writing; R.A.S.: provision of study material or patients; A.I.C.: conception/design, financial support, provision of study material or patients, data analysis and interpretation, manuscript writing, final approval of manuscript.