Secretion of proteins and antibody fragments from transiently transfected endothelial progenitor cells

Abstract In neurodegenerative diseases such as Alzheimer's disease, Parkinson's disease, multiple sclerosis and amyotrophic lateral sclerosis, neuroinflammation can lead to blood‐brain barrier (BBB) breakdown. After intravenous or intra‐arterial injection into mice, endothelial progenitor cells (EPCs) home to the damaged BBB to promote neurovascular repair. Autologous EPCs transfected to express specific therapeutic proteins offer an innovative therapeutic option. Here, we demonstrate that EPC transfection by electroporation with plasmids encoding the reporter protein GFP or an anti‐β‐amyloid antibody fragment (Fab) leads to secretion of each protein. We also demonstrate the secreted anti‐β‐amyloid Fab protein functions in β‐amyloid aggregate solubilization.

or chemokine secretion. 8 Integration of the EPCs into the compromised BBB promotes neurovascular repair. [9][10][11][12] EPCs can also be transfected ex vivo to produce therapeutic molecules such as antibody fragments (Fabs) directed against misfolded proteins for neuroprotection.
At least 26 clinical trials are registered on www.clini caltr ials. gov using EPCs as therapeutic agents for indications such as ischaemic heart disease, pulmonary arterial therapy and decompensated liver cirrhosis. 13 Autologous EPCs transfected to express specific therapeutic proteins offer an option to treat these and other indications. Transfection can be performed using a number of modalities including several viruses, which can produce longterm expression. Non-viral vectors such as plasmids can be delivered into cells by mechanical, chemical or physical means, primarily producing transient expression since genomic integration does not occur. Electroporation, a physical method, was used in this study because this technique does not require external reagents that may generate unintended consequences.
Anti-β-amyloid monoclonal antibodies (mAbs) for treatment of AD have been tested in several major Phase III clinical trials.
Such antibodies against different epitopes within the biomarker Aβ 1-42 yielded excellent results in vitro and in transgenic mice. [14][15][16][17][18][19][20] Phase I and II clinical trials by AC Immune, 21 Hoffmann-La Roche, 16,22,23 Eli Lilly 17,23,24 and Biogen 19,25,26 have demonstrated encouraging results. However, despite differences in the nature of these antibodies, each failed in Phase III, although later some therapeutic activity was demonstrated with aducanumab (Biogen). The cause(s) of these failures have not been clearly elucidated. Possible causes may be either an increasing lack of therapeutic significance of β-amyloid, which is very unlikely if we take into account the in vitro and in vivo preclinical results, or the fact that the mAbs do not reach their target, the β-amyloid deposits, in a sufficient amount. This is likely due to the strong filtering effect of the BBB. High doses of injected antibodies (60 mg per kg body weight dose) were necessary with aducanumab (Biogen) to obtain a modest effect. 19 We report here a novel targeted delivery system consisting of ex vivo transfected, autologous endothelial precursor cells (EPCs) capable of homing to the BBB and expressing therapeutic Fabs. These cells were obtained for the development of targeted cell-mediated gene therapy to a hypoxic site. 27 EPCs are able to repair the damaged BBB and blood-spinal cord barrier characteristic of neurodegenerative diseases such as AD, 28 ALS, 29,30 traumatic brain injury 31,32 or stroke 33,34 and will thus have a double function when combined with Fab production.

| Plasmids
Commercially available reporter plasmids encode firefly luciferase

TA B L E 1 Electroporation protocols
corresponding to the ubiquitin promoter, a downstream hygromycin resistance marker for cell selection and a polyadenylation sequence. PCR was performed to append optimized peptide signals to both chains of the anti-β-amyloid Fab-encoding sequences (C H1 -V H and C L -V L ). In addition, a 10-His Tag encoding sequence was added to the heavy chain encoding sequence. Finally, the fragments were subcloned into the parent vector with CAG promoters for both chains to create the final vector expressing and secreting the anti-β-amyloid Fab. pSF-CAG.InsSP-GFP and pl.DualCAG.
Hygro.cAb2789 were verified with restriction digests and Sanger sequencing. All plasmids were commercially prepared with endotoxin levels confirm to be <100 EU/mg (Aldevron; OxGene) and diluted to 2 mg/mL in physiological saline.

| Transfection
Electroporation was performed in cuvettes using an ECM 830 (BTX Harvard Apparatus). For simplicity, cells were suspended in culture medium throughout the delivery. The tested pulse protocols were chosen from the literature ( Table 1).

| Reporter assays
Luciferase activity was quantified 20 hours after transfection in medium containing 250 µg/mL luciferin using an Omegastar or

| Viability assays
PrestoBlue (Invitrogen, Thermo Fisher Scientific) in medium was added 20 hours after transfection. After two-hour incubation, reagent reduction was quantified using a FLUOstar Omega or Clariostar microplate reader (BMG Labtech).

| Statistics
The statistical significance between the groups was determined by analysis of variance with Tukey-Kramer multiple comparisons test (GraphPad Software) or by Student's t test. A P value <.05 was considered significant. The mature BBB cell line hCMEC/D3 was also tested using these conditions. In these cells, pulsing with one or two pulses similarly decreased viability ( Figure 1C). In this case, luciferase activity increased nearly threefold with the two-pulse protocol, a rate similar to that of EPCs ( Figure 1D). Subsequent experiments were performed using the two-pulse version of PR0462.

| RE SULTS
A plasmid encoding GFP was designed to generate reporter protein secretion as a model for soluble protein therapies (pSF-CAG.InsSP-GFP, Figure 2A). Transfection efficiency and median cell fluorescence were quantified by flow cytometry and compared between this plasmid and a commercially available GFP plasmid (gWizGFP). Transfection efficiency did not vary significantly cAb2789 using both a chimeric anti-His Tag antibody ( Figure 3D) and a rabbit anti-His Tag antibody ( Figure 3F). Therefore, both mRNA and protein assays confirmed the expression of the transfected Fab.
An aggregate solubilization assay was used to demonstrate both secretion and function of the anti-β-amyloid Fab protein (Figure 4).
Incubation of an irrelevant antibody had no effect on the aggregates.

| D ISCUSS I ON
In this study, we demonstrated that EPCs can be transfected using electroporation at high efficiency with a minor loss of cell viability. Transfected EPCs are capable of secreting a reporter protein. Finally, we show that transfected EPCs can also secrete functional antibody fragments.
We first tested HEPC.CB1 EPC transfection using publicly described pulse protocols. In the early development of electroporation as a delivery technique, an exponential pulse shape, essentially a capacitor discharge, was used for cell transfection. This pulse shape was previously tested in EPCs 38 compared with viral delivery and was found to be unsuitable. In this study, we chose square wave pulse protocols to better control cell exposure to electric pulses.
While delivery with pulse protocol PR0329, originally developed for human umbilical vein endothelial cells (HUVECs), produced significant luciferase expression, it also massively reduced cell viability.
HUVEC viability was not assayed in the original PR0329 protocol.
The reverse was true with protocol 'M', 39 which did not produce detectable cell killing; neither did it produce detectable luciferase expression. It must be noted that protocol 'M' was developed for drug rather than for plasmid delivery to human microvascular endothelial cells. A pulse protocol similar to PR0462 was previously described. 40 HUVECs were suspended in a specific buffer formulation with plasmid and exposed 5ms pulses at a frequency of 1 Hz. While the voltage-to-distance ratio was similar (600 vs. 625 V/cm), a larger number of pulses (8) were delivered. The maximum transfection efficiency observed was 40% with an approximately 10% viability determined using microscopic morphology. Delivery with protocol PR0462 produced some loss of cell viability, but also significant luciferase reporter expression. This single pulse protocol was developed for delivery to HUVECs, but increasing to two pulses increased luciferase levels approximately twofold while maintaining cell viability. This comparison was made in BBB endothelial cells with similar results.
With two pulses, no additional effect on viability was observed, but luciferase levels increased 2.4-fold. This implies that this pulse protocol can reproducibly deliver pDNA across related cell types.
Transfection efficiency was quantified using two different plasmids encoding GFP, a classic reporter plasmid and a plasmid specifically designed for GFP secretion. A transfection efficiency of greater than 80% was observed in HEPC.CB1 EPCs using either of these plasmids. As expected, intracellular GFP was significantly lower in the plasmid designed for secretion than the classic plasmid, while significant levels of GFP were detected in the cell medium.
We next confirmed production and function of secreted anti-β-amyloid Fab after transfection. Production of Fab mRNA was confirmed via qPCR while production of the Fab protein was confirmed microscopically. Function of the secreted anti-β-amyloid Fab was demonstrated using an aggregate solubilization assay. These results support the therapeutic potential of transfected EPCs.
For clinical application of this technology, long-term expression may be desirable for therapeutic efficacy. In that case, viral delivery, particularly with adeno-associated virus (AAV) is unquestionably a promising transfection option. Although AAV integrates preferentially into chromosome 19, random integration may produce insertional mutagenesis. 41 Non-integrating AAV vectors will avoid this problem; however, transgene expression is diluted with cell replication. 42 The purpose of the present study was the demonstration that early EPCs transfected ex vivo with an Fab-encoding plasmid would express and secrete functional Fabs that could solubilize β-amyloid aggregates. The work reported here constitutes the first step toward a complex in vivo study aiming to demonstrate the insertion of injected EPCs into the mouse BBB without damage to the BBB, the penetration of the transfected EPCs and the secretion of the expressed Fabs into the brain parenchyma. This work is underway.

ACK N OWLED G EM ENTS
This work was supported by ALSaTECH Inc, Boston, MA, USA. Writing-review & editing (equal).

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding authors upon reasonable request.