Assessment of biodistribution using mesenchymal stromal cells: Algorithm for study design and challenges in detection methodologies

ARTICLE IN PRESS Cytotherapy, 2017; ■■: ■■–■■

Assessment of biodistribution using mesenchymal stromal cells: Algorithm for study design and challenges in detection methodologies

BLANCA REYES1, MARIA ISABEL COCA1, MARGARITA CODINACH1, MARÍA DOLORES LÓPEZ-LUCAS2, ANNA DEL MAZO-BARBARA1, MARTA CAMINAL1, IRENE OLIVER-VILA1, VALENTÍN CABAÑAS2, SILVIA LOPE-PIEDRAFITA3,4, JOAN GARCÍA-LÓPEZ1,5, JOSÉ M. MORALEDA2, CESAR G. FONTECHA6 & JOAQUIM VIVES1,7,8 1

Servei de Teràpia Cellular, Banc de Sang i Teixits, Barcelona, Spain, 2Unidad de Terapia Celular y Trasplante Hematopoyético, Hospital Clínico UniversitarioVirgen de la Arrixaca, Universidad de Murcia, IMIB, Murcia, Spain, 3 Servei de Ressonància Magnètica Nuclear, Universitat Autònoma de Barcelona, Cerdanyola delVallès, Spain, 4Centro de Investigación Biomédica en Red-Bioingeniería, Biomateriales y Nanomedicina (CIBER-BBN), Universitat Autònoma de Barcelona, Cerdanyola delVallès, Spain, 5Chair of Transfusion Medicine and Cellular and Tissue Therapies, Universitat Autònoma de Barcelona, Bellaterra, Cerdanyola delVallès, Spain, 6Reconstructive Surgery of the Locomotor System,Vall d’Hebron Institut de Recerca (VHIR), Universitat Autònoma de Barcelona, Barcelona, Spain, 7 Departament de Medicina, Universitat Autònoma de Barcelona, Bellaterra, Cerdanyola delVallès, Spain, and 8Tissue Engineering Group,Vall d’Hebron Research Institute (VHIR), Universitat Autònoma de Barcelona, Barcelona, Spain Abstract Background aims. Biodistribution of candidate cell-based therapeutics is a critical safety concern that must be addressed in the preclinical development program. We aimed to design a decision tree based on a series of studies included in actual dossiers approved by competent regulatory authorities, noting that the design, execution and interpretation of pharmacokinetics studies using this type of therapy is not straightforward and presents a challenge for both developers and regulators. Methods. Eight studies were evaluated for the definition of a decision tree, in which mesenchymal stromal cells (MSCs) were administered to mouse, rat and sheep models using diverse routes (local or systemic), cell labeling (chemical or genetic) and detection methodologies (polymerase chain reaction [PCR], immunohistochemistry [IHC], fluorescence bioimaging, and magnetic resonance imaging [MRI]). Moreover, labeling and detection methodologies were compared in terms of cost, throughput, speed, sensitivity and specificity. Results. A decision tree was defined based on the model chosen: (i) small immunodeficient animals receiving heterologous MSC products for assessing biodistribution and other safety aspects and (ii) large animals receiving homologous labeled products; this contributed to gathering data not only on biodistribution but also on pharmacodynamics. PCR emerged as the most convenient technique despite the loss of spatial information on cell distribution that can be further assessed by IHC. Discussion. This work contributes to the standardization in the design of biodistribution studies by improving methods for accurate assessment of safety. The evaluation of different animal models and screening of target organs through a combination of techniques is a cost-effective and timely strategy. Key Words: advanced therapy medicines, animal models, biodistribution, mesenchymal stromal cells, preclinical

Introduction Among current developments in cell-based medicines, multipotent mesenchymal stromal cells (MSCs) account for numerous clinical trials in the fields of regenerative medicine, immunotherapy and organ transplantation [1]. From a regulatory perspective, cell-based medicinal products (CBMPs) must be manufactured according to current Good Manufacturing

Practices, and regulatory authorities evaluate quality and safety data before use in humans [2,3]. However, limited understanding of the complex biology of these products and their behavior in the host pose unprecedented challenges to both regulators and developers for the standardization of methods for cellular characterization and manufacture. Regarding product quality, the International Society for Cellular Therapy has established a minimal criteria

Correspondence: Joaquim Vives, PhD, Servei de Teràpia Cellular, Banc de Sang i Teixits, Edifici Dr. Frederic Duran i Jordà, Passeig Taulat, 116, Barcelona 08005, Spain. E-mail: [email protected] (Received 3 February 2017; accepted 16 June 2017) ISSN 1465-3249 Copyright © 2017 International Society for Cellular Therapy. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jcyt.2017.06.004

ARTICLE IN PRESS 2

B. Reyes et al.

for defining the identity and potency of human MSCs [4]. However, batches of MSCs differ in their characteristics as a result of the (i) source tissue (typically, bone marrow, adipose tissue and umbilical cord), (ii) donor variability, and (iii) the existing diversity of procedures for their isolation, expansion, storage and final formulation [5–8]. Even more critical is the standardization of strategies aiming at evaluating preclinical safety of CBMP, which cannot be addressed by following methods commonly used with small molecule drugs or biopharmaceuticals [9,10]. Scarce instructions and poorly described recommendations in current guidelines lead to a case-by-case assessment in agreement with the competent regulatory authority. Moreover, this type of studies are expected to be conducted in accordance with Good Laboratory Practices [11]. Major safety concerns for CBMP include their (i) biodistribution, which involves tracking, homing and persistence; (ii) tumorigenic potential, or the appearance of genetic abnormalities acquired by the cells during the manufacturing process; and (iii) immunogenicity, despite MSCs being immunoprivileged due to low-level expression of HLA-DR.Tracking cells within the recipient is key to a proper evaluation of their therapeutic effect as well as an objective risk assessment with respect to inappropriate ectopic tissue formation, alteration of the microenvironment at the engraftment site, or tumorigenicity [9,10,12]. Biodistribution patterns of MSCs in the host may differ for each route of administration and condition to treat. If cells are to be administered intravenously, rather than locally, broad dissemination is likely to occur [13,14]. Even after local administration, cells can still migrate, extravase and eventually diffuse to other tissues and can potentially alter their phenotype by responding to either physiologic or pathologic microenvironments [14], and this may lead to unwanted effects. Therefore the understanding of cell persistence, distribution and behavior of the cells after administration are key before clinical use. In the present study, preclinical biodistribution studies supporting current clinical trials of MSCbased products are presented and their advantages critically evaluated. We aimed to define a decision tree based on these data to assist in the design of biodistribution studies and the choice of methodologies for labeling and detection, thus contributing to standardization. Methods Animals All animal care and experimental procedures adhered to the recommendations of local, national and European laws (Decret 214 de 1997, Real Decreto 53 de

2013 and European directive 86/609/CEE of 1986, respectively) and were approved by local Ethical Committees on Human and Animal Experimentation of the Universitat Autònoma de Barcelona’s, Institut de Recerca de l’Hospital de la Vall d’Hebron’s and the Hospital Clínico Universitario Virgen de la Arrixaca, Universidad de Murcia. Animal species and strains used are listed in Table I. Cell therapy product preparation Clinical-grade MSCs derived from either bone marrow (BM) or Wharton’s jelly (WJ) were produced within the context of five clinical trials (ClinicalTrials .gov identifiers NCT01227694, NCT01605383, NCT02630836, NCT02566655 and NCT03003364) with appropriate donor-informed consent and were further expanded in vitro by using Dulbecco’s Modified Eagle’s Medium (Gibco) supplemented with 10–20% hSerB (Banc de Sang i Teixits) containing 2 mmol/L glutamine [15,17,18]. Fucosylated BM-MSCs were generated from samples taken from three volunteers (ClinicalTrials.gov identifier NCT02566655) with appropriate donorinformed consent and approval by the local ethics committee. Briefly, cells were seeded in low-glucose α-MEM 94% (Gibco) supplemented with 5% human platelet lysate (hPL; Centro Regional de Hemodonación de Murcia), 50 U/mL penicillin/streptomycin (Lonza Biologics) and 2 U/mL heparin (Mayne Pharma) using TrypLE Select (Gibco) for lifting cells in each culture passage [19]. After the third passage, exo-fucosylation was performed using α-(1,3)-fucosyltransferases VI (FTVI) (R&D Systems) and 1 mmol/L GDP-fucose as substrate (Sigma-Aldrich). Reactions were performed at 37°C for 1 h with gentle shaking followed by centrifugation and washing with Dulbecco’s phosphatebuffered saline (PBS; Gibco) [20]. Ovine MSCs were derived either from BM aspirates [21] or from amniotic fluid as briefly described next. Ovine amniotic fluid MSCs (oAF-MSC) cultures were started with the isolation of plasticadherent cells followed by a cell expansion step. In the isolation stage, 300 cm2 total surface culture area (Corning) were seeded with 0.27 mL oAF per square centimeter in M199 (Gibco, Invitrogen) supplemented with 10% v/v fetal bovine serum (Biological Industries), 5 ng/mL basic Fibroblast Growth Factor (bFGF; Cellgenix) and 1 × penicillin/streptomycin/ amphotericin B (Gibco, Invitrogen). Non-adherent cells and debris were removed by washing with PBS (Thermo Fisher Scientific Hyclone) after 3 days of culture and adherent cells cultured with fresh growth medium that was subsequently changed twice a week. After 10 days in culture, cells were harvested using 0.05% trypsin-EDTA dissociation solution (Gibco,

Table I. Summary of biodistribution studies. Animal model description Clinicaltrial/ disease model

Systemically infused eGFP- hMSCs in immunodeficient mice

Mus musculus (NRG, n = 10 female)

4 × 105

3 mo

Intra-articularly infused eGFPhMSCs in athymic rats

Rattus norvegicus (RNU, n = 6 male)

6 × 105 6 × 104

1 mo

Intrathecally infused hUC-MSCs in athymic rats Intravenously infused hUC-MSC in athymic ratsa

Rattus norvegicus (RNU, n = 9 male) Rattus norvegicus (RNU, n = 6 male)

1 × 106

3 mo

1 × 106

3 mo

Subdermally transplanted hMSC+bone+fibrin constructs in athymic mice Systemically infused fucosylated hMSC in immunodeficient mice

Mus musculus (NIH nude, n = 6 female)

1 × 106

2 mo

Mus musculus (NOD/SCID, n = 5 female and n = 6 male)

1 × 106

MMC

Cell-fate tracking of MPIO-labeled AF-oMSC+fibrin administered to MMC fetuses

Ovis aries (Ripollesa breed, n = 4 female and male); surgically induced MMC

NCT01605383 ONFH

Cell-fate tracking of eGFP-labelled oMSC+bone+fibrin administered to osteonecrotic femoral headb

Ovis aries (Ripollesa breed, n = 1 female); surgically induced ONFH

NCT01227694 Osteoarthrosis

NCT03003364 Spinal Cord Injury

NCT02630836 Femur fractures

NCT02566655 Osteoporosis

Dose

Follow-up

Findings

Technique

Marker

Tissues

Comments

eGFP-labeled cells were found in liver and spleen; human genetic material detected in kidneys and lungs IHC demonstrated presence of labeled cells in infrapatellar fat pad; no genetic material found in any major organ No genetic material was found in major organs

PCR Bioimaging IHC

eGFP eGFP hMitochondria

Lung, liver, gonads, spleen

Robust by combining results from 3 techniques

PCR IHC

eGFP eGFP hMitochondria

Lung, liver, gonads, kidneys, spleen

Results dependant on quality of genomic DNA extracts

PCR

CART1

Persistence of UC-MSCs preferentially in the spleen, although presence of human genetic material was also detected in lungs, liver and kidney No human genetic material found in any major organ

PCR

CART1

PCR

CART1

Lung, liver, gonads, kidneys, spleen

Results dependant on the quality of genomic DNA extracts

24 h, 48 h; 1, 5, 7, 12 wk

Human genetic material found in lungs and spleen (β-actin), ovaries and brain (β2-microglobulin)

Real-time PCR

Human β2-microglobulin human β-actin

High background but detection of genetic material from living cells

4 × 106

11, 37, 49 d

Labeled cells persisted in the administration site

MRI, Perl’s Prussian blue staining

MPIO

Lungs, liver, gonads, kidneys, spleen, heart, brain, blood, bone marrow Lamb vertebrae

6 × 106

3 mo

Labeled cells persisted in the administration site

IHC

IHC

Femoral head

Off-target labeling of macrophages after phagocytosis of labeled MSC; interference from blood cells Time-consuming; low throughput

3

List of biodistribution studies performed, including the experimental system, type of cells and methodology used for labeling and detection. AF, amniotic fluid; hMSC, human MSCs; MMC, myelomeningocele; N/A, not applicable; NIH, U.S. National Institutes of Health; NRG, NOD.Cg-Rag1tm1Mom Il2rgtm1Wjl/SzJ; NOD/SCID, non-obese/ severe combined immune deficiency; ONFH, osteonecrosis of the femoral head; oMSC, ovine multipotent MSCs; UC, umbilical cord. Cases in which partial data from the studies has been presented elsewhere are indicated as follows: areference [15]; breference [16].

ARTICLE IN PRESS

Experimental system (n)

Decision tree for the design of MSC biodistribution studies

Study description and route

Methodologies

ARTICLE IN PRESS 4

B. Reyes et al.

Invitrogen). Reseeding was performed at 3 × 103 viable cells/cm2 in new flasks and cultured for 6–7 additional days. All cultures were maintained at 37°C and 5% CO2 in humidified incubators. Labeling Lentiviral transduction of MSCs: Semi-confluent MSC cultures were transduced with lentiviral vectors encoding the enhanced green fluorescent protein (eGFP) under the control of the internal spleen focus forming virus promoter at multiplicity of infection (MOI) of 250, as reported elsewhere [16]. The percentages of eGFP-positive cells were determined by flow cytometry using a FACSCalibur device (Becton Dickinson). MPIO cell labeling: oAF-MSCs were labeled with dragon green fluorescent 1.63 micrometer-sized superparamagnetic iron oxide particles (MPIO; Bangs Laboratories) at 100 MPIO:1 viable cell ratio. After 20–24 h, internalization of MPIO in cells was verified by visualization under a fluorescence microscope (Nikon). The percentage of labeled cells was determined by flow cytometry (FACSCalibur, BD) after trypsinization and counting. Detection Animals were euthanized with an overdose of sodium pentobarbital (Vetoquinol) administered intraperitoneally in rodents or intravenously in sheep. Bioimaging Major organs were harvested and eGFP expression was measured ex vivo by using an IVIS Spectrum and analyzed with the Living Image software (Caliper Life Sciences). Polymerase chain reaction (PCR) Two modalities of PCR were investigated, endpoint PCR and real-time quantitative PCR (qPCR). For endpoint PCR, Genomic DNA was extracted using the AllPrep DNA/RNA/Protein Mini Kit (Qiagen) according to the manufacturer’s instructions. Absorbance readings were performed at 260 (A260) and 280 nm (A280) in a NanoDrop Lite spectrophotometer (Thermo Fisher Scientific). DNA samples were subjected to amplification using HotStar Taq polymerase (Qiagen) with specific primers according to a method published elsewhere [22]. Primers used for amplification were as follows: human CART1 (156-bp fragment): AAGGATACCAC AATAAGCTGC (forward) and GGTTTGTGGAGA CTGGCAC (reverse); rat GAPDH (489-bp fragment): CAAAGCACCTTCAAGTGCCC (forward) and AGCCCTCCCTTCTCTCGAAT (reverse); eGFP (300-bp fragment): CCTACGGCGTGCAGT

GCTTCAGC (forward) and CGGCGAGCTGCACG CTGCGTCCTC (reverse); mouse β-actin (1000-bp fragment): GATGACGATATCGCTGCGCTGG TCG (forward) and GCCTGTGGTACGACCA GAGGCATACAG (reverse). For real-time qPCR, frozen tissues (including heart, lungs, liver, spleen, kidney, gonads, central nervous system and peripheral blood) were disrupted using the TissueLyser LT device (Qiagen), and RNA was extracted using RNeasy and RNeasy Plus Mini kits (Qiagen). The High Capacity cDNA Reverse Transcription Kit (Qiagen) was used for the synthesis of cDNA, and probes were as follows: mouse β2microglobulin [Mm_B2m_1_FAM QuantiFast], human β2-microglobulin [Hs_B2M_2_FAM QuantiFast], mouse β-actin [Mm_ACTB_2_FAM QuantiFast], human β-actin [Hs_ACTB_2_FAM QuantiFast] (all from Qiagen).The QuantiFast Multiplex PCR Kit was used for the amplification in a RotorGene device (Qiagen). Cell tracking by magnetic resonance imaging (MRI) MRI was performed using a 7-T Bruker BioSpec spectrometer (Bruker BioSpin). Lamb vertebrae were embedded in 1% agarose and placed into a volume coil with an inner diameter of 72 mm. T2-weighted images were acquired in the transverse plane with a fast spin echo sequence at a 0.14 × 0.15 × 1 mm3 spatial resolution (repetition time = 4 s; echo time = 20 ms; echo train length = 4; matrix size = 512 × 512). Histological processing and immunohistochemical procedures Tissues were fixed in 4% (v/v) paraformaldehyde solution (Sigma-Aldrich) at room temperature for 2 weeks, decalcified with 5% formic acid (Decalcifier I, Surgipath Canada) for 4–6 weeks in the case of bone samples, and embedded in paraffin for subsequent histological and immunohistochemical procedures. Microtome sections 4 µm thick were stained with hematoxylin and eosin (Sigma-Aldrich). Grafted cells were detected immunohistochemically using either polyclonal antibodies against the eGFP reporter protein (GFP10-20; Aves Labs) or mouse monoclonal antibodies against human mitochondria (113-1; EMD Millipore) labeled using the ARK (Animal Research Kit, Dako). Briefly, tissue sections were de-paraffined, hydrated and treated with 3% (v/v) hydrogen peroxide (Sigma) for 30 min at room temperature to block endogenous peroxidase activity. Heat-induced antigen retrieval was performed in 0.01 mol/L sodium citrate at pH 6.0 (Sigma) and 10% (v/v) bovine serum albumin (Sigma) in PBS at room temperature for 30 min to block nonspecific antibody-binding sites. Sections were then incubated with the primary antibody (dilution 1:200) overnight at 4°C. Biotinylated secondary rabbit

ARTICLE IN PRESS Decision tree for the design of MSC biodistribution studies anti-chicken immunoglobulin Y antibodies (ab6752; Abcam) were incubated for 1 h at dilution 1:200 at room temperature and antibody binding was visualized by using Universal LSAB2 Kits (Dako) in combination with diaminobenzidine substrate and counterstained with either hematoxylin (SigmaAldrich) or methyl green (Takara Biomedicals).

5

Data analysis Descriptive data were expressed as mean ± SD (number of replicates). Results and discussion Decision tree

Staining of iron nanoparticles Prussian blue staining was used for the detection of cells loaded with iron nanoparticles. Briefly, tissue sections were de-paraffined and hydrated, incubated with 10% potassium ferrocyanide (Perls reagent, Dako) in 10% hydrochloric acid (Sigma) for 30 min, washed and counter-stained with Nuclear fast Red solution (Dako).

The analysis of biodistribution studies using MSCs presented in Table I allowed identification of the critical points in the design of this type of preclinical pharmacokinetics study (Figure 1), which were then grouped into four main categories: (i) animal model and route of administration, (ii) heterologous or homologous nature of the cell therapy product, (iii) labeling, and (iv) methodologies for detection.

Figure 1. Decision tree algorithm for the design of biodistribution studies. Critical points: (I) animal model selection (small or large) and route of administration (systemically or intended), (II) nature of cellular product (heterologous or homologous), (III) labeling (nonrequired, genetic or chemical) and follow-up (short or long term), and (IV) methodologies for detection: nucleic acid amplification (endpoint PCR or qPCR); IHC, bioimaging and MRI. BioD, biodistribution; PoC, proof of concept.

ARTICLE IN PRESS 6

B. Reyes et al.

Animal model and route of administration The first critical point considered was the choice of the animal model and the route of administration (Figure 1). Overall, working with small animal models, commonly rodents such as mice and rats, is cheaper than working with larger models, and systemic administration is typically used in biodistribution studies because it is considered to represent the worst-case scenario because of the high level of organ exposure to the cell therapy product. The use of larger animals is more expensive because it incurs high maintenance costs and requires clinical equipment; however, they better mimic human morphology and physiology, so administration can be done via the same route as intended in humans and allows for the assessment of efficacy endpoints, in addition to persistence and biodistribution of MSCs [23]. In our case, ovine models of myelomeningocele and osteonecrosis of the femoral head were used for determining the persistence of the cellular product in the target organ (lamb vertebrae and proximal femur, respectively) in a “proof of concept” setting that yielded crucial data for establishing starting doses in humans (Table I). For the regulatory approval of initial phases of clinical development, biodistribution studies using small animals administered systemically and by the intended route are accepted by the competent regulatory authority. An example of studies not reproducing the pathological condition in humans, but using the intended route of administration for the evaluation of biodistribution in principal organs included intrathecal injection of MSCs in athymic rats (Table I). Heterologous or homologous nature of the cell therapy product The selection of the animal model and the route of administration influence the decision of using either heterologous or homologous cell therapy products (Figure 1), which in turn determine the labeling technique for the assessment of biodistribution. Small animal models allow the administration of the original heterologous product of human origin, ideally manufactured under Good Manufacturing Practice conditions, without risk of immunorejection by using immunocompromised strains. On the contrary, larger animals involve the manufacturing of specific homologous cell therapy products, ideally in a Good Laboratory Practices–compliant setting, which in our case were ovine MSCs isolated from either BM or amniotic fluid (Table I). Labeling Labeling can be circumvented in the case of heterologous or xenotransplantation of CBMP in which the

tracking of the cellular product in the recipient is based on species-specific differences. In this case, we amplified human CART1 and β2-microglobulin in murine tissue specimens. Lack of labeling for homologous products is only found in the allogeneic context in which detection can be based on polymorphisms or sex-chromosome analyses. Labeling can be classified either as genetic or chemical, and the election of either determines the follow-up methodologies (Figure 1). For long-term studies, genetic labeling is required, because it is maintained in daughter cells, unlike chemical labeling that is diluted on cell division. Genetic engineering of stem cells is commonly used for the generation of cell lines expressing reporters such as luciferase [13], β-galactosidase/LacZ [24] and eGFP [25]. Potential disadvantages of genome editing include the possibility of altering genetic stability, phenotype and multipotentiality of cells, in addition to the need for protocol optimization, for high transducing efficiencies, and the time restrictions between sample collection and in vivo use (particularly in the autologous setting), that may prevent from sorting, and thus enriching, the labeled population. As an example, we experienced inefficient transduction of MSCs with retroviruses or lentiviruses encoding the luciferase sequence in a timely manner and thus ultimately opted for labeling human and ovine MSCs with lentiviruses encoding the eGFP reporter at efficiencies of 45.0 ± 10.6% (n = 3) and 66.0 ± 22.5% (n = 3), respectively (Figure 2A). oMSCs labeled using MPIO were administered to a myelomeningocele model and cell persistence assessed by MRI, yielding 93.2 ± 2.6% (n = 4) labeling efficiencies without altering the cellular morphology (spindle-shaped in culture), viability (>95% after trypsinization) or kinetic parameters (3.3 ± 1.3 cell doublings in 7 days of culture versus 3.2 ± 0.5 CPD in non-labeled MSCs) (Figure 2D,E). Methodologies for detection The final critical point to consider is the choice of the detection method that must be consistent with the labeling of the cell therapy product, if applicable (Figure 1). In the present study, the methodologies used included: PCR, qPCR, IHC, bioimaging and MRI. Amplification of nucleic acids PCR is commonly used for evaluating biodistribution patterns by amplifying sequences of plasmids from DNA vaccines and gene therapy [26]. In the development of CBMP, PCR has many advantages to offer—namely cost, time, throughput and simplicity (Figure 1). However, high-quality intact nucleic acids from target tissues is the first and most critical

ARTICLE IN PRESS Decision tree for the design of MSC biodistribution studies

7

Figure 2. Challenges in the labeling of MSCs, processing of samples and analysis of cellular residence in target tissues. (A) Decay of eGFP expression along passage number in two ovine MSC lines (● and ■). (B) The Quality of DNA extracted using standard DNA extraction kits varied widely depending on the nature of each tissue. (C) Sensitivity of endpoint semi-quantitative PCR with regard to the number of eGFP-labeled MSCs (at 51.75% transduction efficiency). (D) T2-weighted MRI scan of a repaired vertebrae in which signal drop associated to MPIO-labeled AF-oMSCs is depicted in the repaired region as a dark hypointense area (black arrows). (E) Similar hypo intensities were detected in hemorrhagic regions thus masking the presence of intact labeled MSCs (white arrow) and blue color after PERL’s staining (inset). (F) Histologic section of a femoral head that was decalcified and paraffinized before detecting eGFP-labeled after IHC procedures with antibodies specific to GFP. (G) A human MSC was detected in the liver of a mouse by using anti-human mitochondria antibodies (nuclei were counterstained with FastGreen). (H) Sensitivity of eGFP detection by bioimaging and (I) presence of eGFP in livers from mice treated with eGFP-labeled human MSCs. Scale bars = 100 µm. FLI, fluorescence imaging.

ARTICLE IN PRESS 8

B. Reyes et al.

requirement. Indeed, we found that both yields and quality of genomic DNA varied widely depending on the physical and biochemical nature of the tissues (Figure 2B). For example, DNA extraction methods may work equally well for tissues such as liver, but fibrous tissues such as heart, fatty tissues such as brain and nuclease-rich tissues like spleen presented challenges in DNA isolation. Genomic DNA extracts from liver and spleen were the ones with the highest purity (A260 = 1.88 ± 0.13% and 1.89 ± 0.02%, respectively; n = 6) when using standard commercial DNA extraction kits, as opposed to preparations from lungs (A260 = 2.10 ± 0.10%; n = 6), despite all samples being processed simultaneously. This may affect the PCR results, particularly because of the need to detect a defined, single band that can be masked in lowquality samples or even lead to false negatives by inhibition of the amplification reaction. To avoid the latter problem, we included internal controls in the same reaction mixture (i.e., β-actin in mouse, GAPDH in rat) by defining multiplex strategies for simultaneous detection of the species-specific internal control and the human gene of interest. One disadvantage that also needs to be acknowledged is that this procedure is invasive; therefore, it is only possible to perform at termination of the in vivo study or, in some cases, on biopsies of a particular tissue, if available. In our hands, sensitivity for detecting human DNA using the standard endpoint PCR was a sample of 100 MSCs (Figure 2C). Alternatively quantitative data can be obtained by qPCR and, although technically more demanding, the use of RNA instead of genomic DNA avoids detection of genetic material from dead cells and provides information on the copy number of the transcripts, which can be compared against controls [27]. We successfully isolated RNA from a panel of organs harvested from animals receiving either fucosylated or unmanipulated MSCs (Table I). Murine sequences encoding for β-actin and β2-microglobulin were successfully amplified in all samples as internal controls thus mitigating the risk of false negative results. The advantages of this technique coincide with those of endpoint PCR (improved sensitivity and allowing for accurate quantification) but incurring higher costs. Staining of histological sections Microscopic examination of tissue samples is a must in any toxicological study. Particularly in the assessment of biodistribution, special attention is given to findings related to ectopic tissue formation or tissue overgrowth. Therefore, histopathology and biodistribution assessment can be performed in combination. However, the identification of a small number of cells at an ectopic location or persisting longer than expected does not in itself mean a halt

CBMP development but identifies an issue that deserves further analysis. In our studies, heterologous MSCs were readily detected at the single cell level by immunochemical staining on histological sections of murine tissue samples that were initially embedded in paraffin by using antibodies that specifically recognized either human mitochondria or the eGFP reporter protein (Table I, Figure 2F,G). Despite using cells labeled with fluorescent reporters, they could not be visualized directly after tissue processing and paraffin embedding; this was also due to the high autofluorescence of the tissue of interest (i.e., bone). Clearly, IHC is a powerful technique but laborious and requires individualized processing of certain tissues (i.e., decalcification of bone) and optimization of protocols (i.e., testing of antigen retrieval methods). Even when optimal detection conditions are well defined, one important point to consider is how representative the analysis of a limited number of sections is with respect to the total size of that organ. Moreover, the silencing of the reporter gene has to be taken into consideration to avoid false negatives as happens with extensive cell division in MSCs (Figure 2A). Bioimaging Whole body imaging capture allows for longitudinal follow-up study designs avoiding interim sacrifice of experimental groups, and thus each animal becomes its own control [13]. This strategy requires the labeling of cells, typically by genetic engineering. Although a number of technologies are currently available (i.e., use of nonviral methods and adenoviral, retroviral and lentiviral gene transfer systems), the success of transduction approaches is largely empirical. However, some determinants are (i) the quality of the preparations of viral particles, (ii) the adaptation of homemade protocols to meet good scientific practices, (iii) the availability of specific cellular samples with appropriate identity and potency, (iv) presence of virus receptors and (v) time restrictions until administration of the MSC-based product into animals. We used eGFPlabeled MSCs requiring ex vivo bioimaging for proper detection due to low tissue penetration of those reported. However, autofluorescence adds background noise. In our hands, eGFP-labeled oMSCs used for ex vivo organ bioimaging allowed for detection greater than 25 × 103 cells (Figure 2H,I).These methodologies provide many advantages, including longitudinal follow-up, whole body analysis and high specificity, but their development requires an accurate setup and highly skilled staff. MRI Labeling of MSCs with MPIOs and subsequent detection by MRI allows for noninvasive monitoring with a low detection limit [28,29]. However, in vivo

ARTICLE IN PRESS Decision tree for the design of MSC biodistribution studies detection in large animals is difficult to perform with standard research equipment, which is better suited for mice or rats [30]. To overcome this issue, we used MRI at termination after cutting sections of the tissue of interest and readily detected MPIO labeled AF-oMSCs in the lamb vertebrae as hypointensity in T2-weighted images (Figure 2D). However, ex vivo observations of the tissue containing labeled AFoMSCs by MRI may be misinterpreted because of the interference of hemorrhagic areas, which also showed a loss of signal in a region where macrophages loaded with hemosiderinic pigment were also detected as positive for Prussian blue staining (Figure 2E). This finding highlights that the identification of labeled AF-oMSCs in the target tissue requires additional specific staining protocols on tissue sections. Hence, MRI showed better sensitivity and tissue penetration than bioimaging, but the signal was detected only temporarily because of dilution of the contrast effect on cell division. We conclude that, from all methodologies tested in the present work, PCR was cost-effective, allowed processing of several samples in the same run and displayed specific detection and adequate sensibility.The combination of PCR and IHC analyses is recommended to gain spatial positioning information in positive samples resulting from PCR analyses. Finally, the design, execution and interpretation of pharmacokinetics studies using MSCs can benefit from the decision tree proposed here. Acknowledgments The authors acknowledge Sandra Barbosa, Eva Cepeda and PatriVergara (Universitat Autònoma de Barcelona, Bellaterra, Spain) and Marta Rosal, Alex Rojo, Cristina León and Eva Morón (Hospital Universitari de la Vall d’Hebron, Barcelona, Spain) for their careful assistance with animal management;Yolanda Fernández, Anna Pujol, Cleofé Romagosa and Rosa M. Rabanal for technical assistance and advice; and Laura BatlleMorera and Clémentine Mirabel for critical review of the manuscript. This work was developed in the context of ADVANCE(CAT) with the support of ACCIÓ (Catalonia Trade & Investment; Generalitat de Catalunya) and the European Community under the Catalonian ERDF operational program (European Regional Development Fund) 2014–2020; and by the Spanish Cell Therapy Network (TerCel, expedient number: RD16/0011/0028, RD16/0011/0001, Instituto de Salud Carlos III, FEDER, European Regional Development Fund 2014-1020, European Union, “potenciar la investigación el desarrrollo tencologia y la innovación,” “una manera de hacer Europa”).This work was also supported by Mutua Madrileña (Expedient No. 08/2749, from 2008 to 2011).

9

Disclosure of interests: The authors have no commercial, proprietary, or financial interest in the products or companies described in this article. References [1] Prockop DJ. The exciting prospects of new therapies with mesenchymal stromal cells. Cytotherapy 2017;19(1):1–8. [2] Commission directive 2003/94/EC of October 2003 laying down the principles and guidelines of good manufacturing practice in respect of medicinal products for human use and investigational medicinal products for human use, 2003. [3] EMA/CAT/486831/2008 Guideline on the minimum quality and non-clinical data for certification of advanced therapy medicinal products, 2008. [4] Dominici M, Le Blanc K, Mueller I, Slaper-Cortenbach I, Marini F, Krause D, et al. Minimal criteria for defining multipotent mesenchymal stromal cells. The International Society for Cellular Therapy position statement. Cytotherapy 2006;8(4):315–17. [5] Wagner W, Wein F, Seckinger A, Frankhauser M, Wirkner U, Krause U, et al. Comparative characteristics of mesenchymal stem cells from human bone marrow, adipose tissue, and umbilical cord blood. Exp Hematol 2005;33(11):1402– 16. [6] Hass R, Kasper C, Böhm S, Jacobs R. Different populations and sources of human mesenchymal stem cells (MSC): a comparison of adult and neonatal tissue-derived MSC. Cell Commun Signal 2011;9:12. [7] Sousa BR, Parreira RC, Fonseca EA, Amaya MJ, Tonelli FM, Lacerda SM, et al. Human adult stem cells from diverse origins: an overview from multiparametric immunophenotyping to clinical applications. Cytometry A 2014;85(1):43–77. [8] Phinney DG, Sensebe L. Mesenchymal stromal cells: misconceptions and evolving concepts. Cytotherapy 2013; 15(2):140–5. [9] Goldring CE, Duffy PA, Benvenisty N, Andrews PW, Ben-David U, Eakins R, et al. Assessing the safety of stem cell therapeutics. Cell Stem Cell 2011;8(6):618–28. [10] Sharpe ME, Morton D, Rossi A. Nonclinical safety strategies for stem cell therapies. Toxicol Appl Pharmacol 2012;262(3):223–31. [11] Organization for Economic Cooperation and Development. OECD principles on good laboratory practice. Paris: OECD Publishing; 2003. [12] Vives J, Blanco M, Caminal M, Coca M, Codinach M, Coll R, et al. Development of an advanced cell therapy product indicated for the treatment of gonarthrosis. BMC Proc 2015;9(Suppl. 9):O9. [13] Vilalta M, Degano IR, Bago J, Gould D, Santos M, Garcia-Arranz M, et al. Biodistribution, long-term survival, and safety of human adipose tissue-derived mesenchymal stem cells transplanted in nude mice by high sensitivity non-invasive bioluminescence imaging. Stem Cells Dev 2008;17(5):993– 1003. [14] Leibacher J, Henschler R. Biodistribution, migration and homing of systemically applied mesenchymal stem/stromal cells. Stem Cell Res Ther 2016;7:7. [15] Oliver-Vila I, Coca MI, Grau-Vorster M, Pujals-Fonts N, Caminal M, Casamayor-Genesca A, et al. Evaluation of a cell-banking strategy for the production of clinical grade mesenchymal stromal cells from Wharton’s jelly. Cytotherapy 2016;18(1):25–35. [16] Caminal M, Velez R, Rabanal RM, Vivas D, Batlle-Morera L, Aguirre M, et al. A reproducible method for the isolation and expansion of ovine mesenchymal stromal cells from bone

ARTICLE IN PRESS 10

[17]

[18]

[19]

[20]

[21]

[22]

B. Reyes et al. marrow for use in regenerative medicine preclinical studies. J Tissue Eng Regen Med 2016;doi:10.1002/term.2254. published online November 18. Codinach M, Blanco M, Ortega I, Lloret M, Reales L, Coca MI, et al. Design and validation of a consistent and reproducible manufacture process for the production of clinical-grade bone marrow-derived multipotent mesenchymal stromal cells. Cytotherapy 2016;18(9):1197–208. Soler R, Orozco L, Munar A, Huguet M, Lopez R, Vives J, et al. Final results of a phase I–II trial using ex vivo expanded autologous Mesenchymal Stromal Cells for the treatment of osteoarthritis of the knee confirming safety and suggesting cartilage regeneration. Knee 2016;23(4):647–54. Carrancio S, Lopez-Holgado N, Sanchez-Guijo FM, Villaron E, Barbado V, Tabera S, et al. Optimization of mesenchymal stem cell expansion procedures by cell separation and culture conditions modification. Exp Hematol 2008;36(8):1014–21. Sackstein R, Merzaban JS, Cain DW, Dagia NM, Spencer JA, Lin CP, et al. Ex vivo glycan engineering of CD44 programs human multipotent mesenchymal stromal cell trafficking to bone. Nat Med 2008;14(2):181–7. Caminal M, Fonseca C, Peris D, Moll X, Rabanal RM, Barrachina J, et al. Use of a chronic model of articular cartilage and meniscal injury for the assessment of long-term effects after autologous mesenchymal stromal cell treatment in sheep. N Biotechnol 2014;31(5):492–8. Murdoch B, Gallacher L, Awaraji C, Hess DA, Keeney M, Jay K, et al. Circulating hematopoietic stem cells serve as novel targets for in utero gene therapy. FASEB J 2001;15(9):1628– 30.

[23] Vives J, Oliver-Vila I, Pla A. Quality compliance in the shift from cell transplantation to cell therapy in non-pharma environments. Cytotherapy 2015;17(8):1009–14. [24] Leiker M, Suzuki G, Iyer VS, Canty JM Jr, Lee T. Assessment of a nuclear affinity labeling method for tracking implanted mesenchymal stem cells. Cell Transplant 2008;17(8):911– 22. [25] van Vollenstee FA, Jackson C, Hoffmann D, Potgieter M, Durandt C, Pepper MS. Human adipose derived mesenchymal stromal cells transduced with GFP lentiviral vectors: assessment of immunophenotype and differentiation capacity in vitro. Cytotechnology 2016;68(5):2049–60. [26] Gonin P, Gaillard C. Gene transfer vector biodistribution: pivotal safety studies in clinical gene therapy development. Gene Ther 2004;11(Suppl. 1):S98–108. [27] Delenda C, Gaillard C. Real-time quantitative PCR for the design of lentiviral vector analytical assays. Gene Ther 2005; 12(Suppl. 1):S36–50. [28] Shapiro EM, Sharer K, Skrtic S, Koretsky AP. In vivo detection of single cells by MRI. Magn Reson Med 2006; 55(2):242–9. [29] Nohroudi K, Arnhold S, Berhorn T, Addicks K, Hoehn M, Himmelreich U. In vivo MRI stem cell tracking requires balancing of detection limit and cell viability. Cell Transplant 2010;19(4):431–41. [30] Jelicks JLA, Tanowitz HB, Peters VM, Mendez-Otero R, Campos de Carvalho AC, et al. Molecular imaging, biodistribution and efficacy of mesenchymal bone marrow cell therapy in a mouse model of Chagas disease. Microbes Infect 2014;16(11):923–35.