Enabling Technologies | Published:

Bioprocess development for canine adenovirus type 2 vectors

Gene Therapy volume 20, pages 353360 (2013) | Download Citation


Canine adenovirus type 2 (CAV-2) vectors overcome many of the clinical immunogenic concerns related to vectors derived from human adenoviruses (AdVs). In addition, CAV-2 vectors preferentially transduce neurons with an efficient traffic via axons to afferent regions when injected into the brain. To meet the need for preclinical and possibly clinical uses, scalable and robust production processes are required. CAV-2 vectors are currently produced in E1-transcomplementing dog kidney (DK) cells, which might raise obstacles in regulatory approval for clinical grade material production. In this study, a GMP-compliant bioprocess was developed. An MDCK-E1 cell line, developed by our group, was grown in scalable stirred tank bioreactors, using serum-free medium, and used to produce CAV-2 vectors that were afterwards purified using column chromatographic steps. Vectors produced in MDCK-E1 cells were identical to those produced in DK cells as assessed by SDS-PAGE and dynamic light scatering measurements (diameter and Zeta potential). Productivities of 109 infectious particles (IP) ml−1 and 2 × 103 IP per cell were possible. A downstream process using technologies transferable to process scales was developed, yielding 63% global recovery. The total particles to IP ratio in the purified product (<20:1) was within the limits specified by the regulatory authorities for AdV vectors. These results constitute a step toward a scalable process for CAV-2 vector production compliant with clinical material specifications.


Viral vectors are currently the most efficient tools for in vivo gene transfer. Owing in part to their in vivo efficiency, adenovirus vectors (AdV) are used more often than any other vector in clinical trials (http://www.wiley.com/legacy/wileychi/genmed/clinical/). Several characteristics make this diverse family of vectors attractive, namely wide cell tropism in quiescent and non-quiescent cells, the poor ability to integrate the host genome and the high production titers obtained in culture.1 However, gene transfer efficacy and the clinical use of human AdV can be hampered by the preexisting humoral and cellular immunity in most humans.2, 3

Canine adenovirus type 2 (CAdV-2, or more commonly referred to as CAV-2) is probably the best-characterized nonhuman adenovirus vector.2, 3, 4, 5 The paucity of neutralizing antibodies and memory T cells in humans, and efficient gene delivery obtained for E1-deleted (ΔE1) CAV-2 vectors in the central nervous system make these viruses promising tools for human gene therapy.2, 5 The extraordinary ability to preferentially transduce neurons, combined with a remarkable capacity of axonal transport, makes CAV-2 vectors candidates for the treatment of neurodegenerative diseases.6

The need for significant amounts of clinical-grade adenovirus vectors, which in some cases may reach 1013 total particles per patient or 1011 infectious particles (IP) per patient, requires efficient and robust processes for production and purification at large scale compliant with good manufacturing practices (GMP).1 Therefore, current lab-scale protocols must be adapted and transferred to a scalable stirred culture system and chromatographic matrices. Moreover, cells cultured with media containing animal serum should be avoided. The undefined composition and high batch-to-batch variability of serum, together with its potential source of contaminations, raises safety concerns and hinders the standardization of cell culture processes for the production of biopharmaceuticals.7

To date, CAV-2 vectors have been produced in a cell line generated from dog kidney (DK).2 However, these cells are not regulated for biopharmaceutical production, which precludes the potential clinical use of CAV-2 vectors. Recently, we developed a CAV-2 E1-transcomplementing cell line based on MDCK. MDCK cells are accepted in Food and Drug Administration and European Medicine Agency, and used for the production of many viral vaccines. In fact, recently, a cell-based flu vaccine produced in MDCK was launched in the market by Novartis Vaccines (Optaflu) (Novartis Vaccines and Diagnostics GmbH & Co. KG, Marburg, Germany). Thus, the use of an MDCK-based cell line would facilitate the regulatory approval for the production of CAV-2 vectors.

Improving vector recovery, minimizing validation requirements and fast-tracking purification are the main downstream challenges for viral vector products.8, 9, 10 The use of chromatographic steps, instead of the traditional CsCl gradient purification, facilitates process scale-up. The currently available chromatographic matrices have pore dimensions (typically 30–80 nm) that exclude viral vectors, restricting their adsorption to the bead surface area, resulting in low binding capacities. This drawback can be circumvented by using tentacle supports, namely membrane chromatography and, more recently, monolithic adsorbent columns.11 These matrices eliminate diffusion limitations, allowing faster volumetric throughput rates increasing speed and productivity. Monolithic adsorbent columns also provide a low-shear fractionation environment that supports high functional recoveries of shear-sensitive viral vectors. Another challenge for virus purification is the improvement of the size exclusion chromatography used for polishing, as it has low loading capacity that limits the scalability of the process and operation at low linear flow rates. Thus, new tools such as matrices operating in negative mode12 are promising alternatives for larger-scale purification of the CAV-2 vectors.

Here, we developed an integrated bioprocess for CAV-2 vector production and purification. MDCK-E1 cells grown in microcarriers were used as host cells for viral production in stirred tank bioreactors, being the virus purification based only on chromatographic steps. The impact of upstream parameters, such as cell inoculum concentration and microcarrier type on cell growth was evaluated; furthermore, the effect of harvesting time on total viral productivity, as well as downstream strategies for clarification, purification and polishing were assessed. Our results constitute a significant step in a scalable CAV-2 vector production and purification process.


The ability of MDCK-E1 cells to produce CAV-2 particles of correct size, charge and SDS profile was evaluated by comparison with viral vector production in DKZeo cells. The impact of serum-free (SF) Optipro medium on cellular growth and viral production was also assessed by comparing the results obtained with serum-supplemented (SS) Minimum Essential Medium Eagle (MEM). Moreover, to develop a scalable process, the best inoculation strategy (cell concentration inoculum and microcarrier type) for improved cell growth and the best time of harvest for maximum virus production yield in stirring conditions were evaluated and confirmed in a 2-l bioreactor. Finally, approaches for vector clarification, purification and polishing were investigated.

MDCK-E1-transcomplementing cells for the production of a ΔE1 CAV-2 vector

To evaluate MDCK-E1 cell line potential for CAV-2 vector production, specific productivity/amplification of CAVGFP and its physical characteristics were investigated. DKZeo cells were used as control because they are currently used for the production of these vectors; vectors were produced in both cell lines using DMEM culture medium.

Similar cell-specific productivities and IP amplification were obtained for both cell lines (Table 1). No significant difference in particle size, charge or protein SDS profile was detected for viral particles produced with DKZeo and MDCK-E1 (Figure 1). These results suggest that MDCK-E1 cells are suitable for CAV-2 vector production.

Table 1: Effect of producer cell line on production and properties of viral particles
Figure 1
Figure 1

(a) SDS-PAGE of purified viral particles produced in DKZeo (lane 2) and MDCK-E1 (lane 3). Gel was loaded with 10 μl of concentrated sample per well. Lane 1: molecular weight markers. (b) Zeta (ζ) potential of purified viral samples produced in DKZeo (, circle) and MDCK-E1 (♦, diamond).

Effect of culture medium on cell growth and vector production under static conditions

The impact of SF medium (Optipro) on cell growth and viral productivity was evaluated by comparison with SS medium (MEM).

Although the specific growth rate of MDCK-E1 cells adapted to SF medium in static conditions was almost half of the value obtained when using SS medium, the maximum cell concentration attained was 3.5-fold higher (1.9 × 106 cells ml−1) (Table 2).

Table 2: Impact of culture medium and culture strategy on specific growth rate (μ) and maximum cell concentration (Xvmax) of MDCK-E1 cells

Cell-specific production of 2223 IP per cell and 391IP per cell was obtained for SF and SS medium, respectively, corresponding to amplification ratios of 445 and 78 IP.

Production optimization under stirring conditions

Effect of cell inoculum and microcarrier type on cell growth

The effect of inoculum concentration on stirred cultures was evaluated in SS and SF medium by testing two concentrations: 1 × 105 cells ml−1 and 2 × 105 cells ml−1. The ability of Cytodex 1 and Cytodex 3 to support cell growth in SF medium was also evaluated.

MDCK-E1 cells did not attach to microcarriers when SF medium was used. To overcome this and promote cell adhesion, 5% (v/v) fetal bovine serum (FBS) was added. After 12 h this culture medium was replaced by regular SF medium (without FBS).

The maximum cell concentration obtained for both culture media and cell inoculums tested is shown in Figure 2. The highest cell concentration was obtained for SF medium at 50 h of culture using an inoculum of 2 × 105 cells ml−1. No differences were observed between the two microcarriers evaluated, as assessed by total cell number and viability. Moreover, a similar range of total cell concentration was obtained for static or stirred cultures using both culture media (Table 2).

Figure 2
Figure 2

Maximum cell concentration obtained in spinner flasks using different cell inoculums in SS and SF media. All experiments were performed with MDCK-E1 and Cytodex 1 except in (♦, diamond) where Cytodex 3 was used. Medium exchange was performed after cell attachment (12 h) and considered to exclude any effect of serum on cell growth under stirring conditions. Error bars represent error associated with the measurement (10%).

Effect of time of harvest on virus yield

After infecting cells, the maximum number of IP was achieved between 24 and 36 hpi (Figure 3), the majority being in the intracellular fraction. At this time, IP amplification ratios of 21 and 28, corresponding to 132 IP per cell and 141 IP per cell, were obtained for MDCK-E1 cells in SF medium and SS medium, respectively.

Figure 3
Figure 3

Amplification ratio of IP along time after infecting MDCK-E1 cells in spinner flasks with SS medium and SF medium. Between 24 and 36 hpi most of the viable cells attained attached to microcarriers. Error bars represent s.d. of triplicate measurements.

Stirred tank bioreactor for ΔE1 CAV-2 production

Taking into account the results obtained in spinner vessels (see sections ‘Effect of cell inoculum and microcarrier type on cell growth’ and ‘Effect of time of harvest on virus yield’), MDCK-E1 cells were cultivated in a 2-l stirred tank bioreactor using an inoculum of 2 × 105 cells ml−1 and 3 mg ml−1 of Cytodex 1 microcarriers. A maximum cell concentration of 1.1 × 106 cells ml−1 was obtained, with the maximum IP productivity being obtained at 24–36 hpi. However, in the bioreactor, an amplification ratio of 256 was attained corresponding to a volumetric productivity of 1.8 × 109 IP ml−1 (Figure 4).

Figure 4
Figure 4

Volumetric productivity (IP ml−1) obtained with MDCK-E1 cultured in static cultures (T-Flasks) and bioreactor (in Cytodex 1) using serum-supplemented medium and SF medium. Viral vectors were collected at 40 and 36 hpi in static and bioreactor cultures, respectively. Error bars represent s.d. of triplicate measurements.

Purification process of CAVGFP

Effect of clarification strategy on viral vector yield

After cell lysis promoted by triton, the bioreactor bulk was treated with 50 U ml−1 of benzonase to digest nucleic acids. Two methods were evaluated for bulk clarification: filtration using depth filters and centrifugation. Depth filters with 0.45 μm pores led to a recovery of 30%. One depth filter with a double layer (0.8+0.45 μm) was introduced after a centrifugation step, and the yield of IP increased to 90% (±2%; n=3).

Effect of flow rate and elution methods during capture/intermediate purification of ΔE1 CAV-2

To determine the best capture/intermediate purification step, two strategies were compared: membrane adsorber (Sartobind Q, Sartorius Stedim Biotech, Goettingen, Germany) and anion exchange monolithic Q (CIM Q) (Bia Separations, Villach, Austria) columns. The effects of increasing flow rate and stepwise elution on recovery were also evaluated, and the efficiency of impurities removal (total protein and DNA) was addressed.

Similar chromatogram profiles were obtained using both strategies, in which three peaks were found. Figure 5a shows the typical chromatogram obtained. However, better resolutions and higher vector yields were achieved using CIM Q (82%±2% recovery, n=3) when compared with Sartobind Q (42%±5% recovery, n=3). Using monolithic column, most of the viral vectors were eluted in the first peak at 21 mS cm−1 (Figure 5a) as confirmed by real-time PCR, electrophoretic analysis and western blot (Figure 5b). The amount of vectors in the other peaks was <7%.

Figure 5
Figure 5

(a) Linear gradient elution profile obtained after loading the 8 ml monolithic column with 120 ml of previously clarified supernatant. Loading and elution were performed at 10 ml min−1. The arrow indicates the peak corresponding to total CAVGFP vectors. (b) SDS-PAGE and western blot analysis samples derived from CAVGFP intermediate purification with anion exchange monolithic column using linear gradient. Bands corresponding to the major viral proteins (II, III and IV) were detected. The polyclonal antibody reacts mostly with one band of molecular weight compatible with penton (protein III) or fiber (protein IV) and low molecular weight protein VI and VII. The gels were loaded with 10 μl of sample per well. 1, molecular weight markers; 2, load of monolithic; 3, flowthrough of monolithic column; 4, peak eluted at 21 mS cm−1; 5, peak eluted at 40 mS cm−1; 6, peak eluted at 65 mS cm−1. (---) Conductivity in mS cm−1; ( − ) absorbance at 280 nm.

The impact of flow rate and elution method was evaluated using the monolithic column (Figure 6). The recovery yield was not affected by the increase in flow rate (up to 40 ml min−1) and the stepwise elution method applied (step increase of NaCl concentration up to 500 mM).

Figure 6
Figure 6

Typical stepwise elution chromatographic profile obtained after loading the monolithic column. The arrow indicates the CAV-2 peak determined by real-time PCR and confirmed with infectious titer assay. The sample was loaded and eluted at 40 ml min−1 and steps of 75, 200, 300 and 500 mM of NaCl were performed. (---) Conductivity in mS cm−1; ( – ) Absorbance at 280 nm.

Therefore, on combining monolithic Q columns with the higher flow rate and step elutions, 89% of protein and 82% of DNA were removed.

Evaluation of two matrices as polishing step

Size exclusion steps using a Superdex 200 column and a core bead prototype resin with mixed properties were evaluated by comparing the recovery yields and dilution factor. The purity/removal of protein and DNA with the best polishing strategy was also assessed.

Figure 7 depicts typical chromatograms obtained from size exclusion chromatography runs. Viral vectors, owing to their very high molecular weight, were excluded from the resin pores, thereby eluting in the void volume of the column. Alternatively, the low-molecular-weight impurities were retained inside the matrix and eluted later as shown in Figure 7a.

Figure 7
Figure 7

Development of a CAV-2 polishing step. (a) Chromatography profiles by size exclusion chromatography after loading 1/10 of the peak collected previously. Two chromatograms are represented. (b) Prototype matrix operating in negative mode. Most of the CAV-2 sample was recovered from the column flow through; the impurities as host cell protein and DNA were retained in the matrix. Black (▪) chromatogram corresponds to an assay where total volume of previous CAV-2 peak was loaded. Gray () chromatogram corresponds to an assay where 1/6 of sample was loaded. (c) SDS-PAGE and western blot of each polishing step strategy. Lanes were loaded with same sample volume. 1, molecular weight markers; 2, sample from size exclusion chromatography peak fraction; 3, sample from flow through of core bead prototype.

Similar recovery yields were achieved using Superdex 200 column or core bead prototype (Table 3). Using this prototype, no dilution of viral samples was observed. On the other hand, a dilution factor of 4–10 was obtained when Superdex 200 column was used.

Table 3: Effect of different purification strategies on ΔE1 CAV-2 yields

The final concentration of impurities was higher in the fraction recovered from Superdex 200 column (Figure 7a) (0.14 mg ml−1 of total protein and 79 ng ml−1 of DNA), as compared with the core bead prototype column (Figure 7b) (0.03 mg ml−1 of total protein and 28 ng ml−1 of DNA). Moreover, a DNA amount of 16 ng per 1010 viral particles was obtained in the fraction collected from core bead prototype. The viral recovery yields were not affected by the flow rates tested (up to 300 cm h−1).

In sum, the overall purification process developed for CAV-2 comprised three main steps: clarification step using centrifugation followed by microfiltration, with 10% losses of viral particles; intermediate purification with a global recovery yield of 82%; and a final polishing step, with a recovery yield of 86% and undiluted sample at the end of process. This procedure can be handled in one working day for scales up to 5-l of bulk material.

The different strategies evaluated are presented in Table 3. The overall process developed had a 63% recovery yield, with an infectious to total particle ratio<20:1.


The potential of CAV-2 vectors for fundamental and clinical gene transfer and the amounts needed require a robust and scalable bioprocess. MDCK cells are accepted by the Food and Drug Administration and European Medicine Agency as producer cells for clinical grade vaccines and recombinant protein. In this study, an MDCK-derived cell line was used to produce ΔE1 CAV-2 vectors using SF medium.

Productivities and the impact on viral particles were consistent with the robustness of MDCK-E1 cells to produce ΔE1 CAV-2 vectors when compared with DKZeo cells.

The main differences in MDCK-E1 cell performance observed between SF Optipro and SS MEM are likely due to the more complex composition of SF medium. Cells cultured in SF medium, although showing slower growth rates, can achieve higher cell concentration as more nutrients and/or nutrient concentration is available. Moreover, a similar range of cell-specific productivity was obtained when using cells cultivated in a richer serum-supplemented medium, namely DMEM.

Adaptation of cells to SF medium affected the cell-specific growth as adherent monolayer in static cultures. However, for stirring conditions similar growth rates were observed, when compared with MDCK-E1 cells growing in SS medium. This observation suggested a relation between agitation and promotion of cell growth rate when SF medium was used.

Serum provides many factors that promote cell adhesion. Although several reports in the literature describe effective cell attachment to microcarriers, for different cell types including MDCK,13 in commercially available serum-free SF formulations, we were not able to accomplish this for the MDCK-E1 cells used herein. Attachment was only significant in the presence of serum, which could be removed afterwards without compromising virus productivity. From a biopharmaceutical production process perspective, the use of animal compounds, namely serum, should be avoided. However, as the serum was only required for the cell attachment phase (inoculation) and was washed out by two complete medium replacement steps, the content of serum during bioprocess cell growth, virus infection and virus production phases was low and the risks associated with its use significantly reduced.

In contrast to the reported cultivation of MDCK cells in a microcarrier under stirring conditions, we observed similar maximum cell concentrations using Cytodex 1 or Cytodex 3 without significant cell detachment during growth.13, 14 In studies using MDCK cells for influenza production, growth rates of about 0.02–0.035 h−1 and maximum cell concentrations of 1.1 × 106–1.3 × 106 cells ml−1 obtained 4 and 5–6 days after inoculation using SS and SF medium, respectively, were reported.13, 15, 16 In our study, the same range of maximum cell concentration was obtained. However, the higher specific growth and shorter lag phase may have been responsible for the maximum cell concentration being attained only 2 days after inoculating the same cell number. Host cell-line effect was excluded, because a similar performance was observed between MDCK-E1 cells and the MDCK progenitor cell line under same conditions (data not shown). These discrepancies may be due to the different culture media used in each study. Moreover, these variations in specific growth rate and maximum cell concentrations should be viewed in the context of the cell metabolic state. From a bioprocess perspective, the need for a higher cell concentration to increase volumetric productivity is expected; thus, the next step toward CAV-2 vector production process is to design a feeding strategy to obtain high MDCK-E1 cell concentration, while maintaining the corresponding cell-specific virus yield.

Although this is a nonhuman adenovirus, the mechanisms underlying viral replication cycle and its production are similar.2, 6, 17 Therefore, a high MOI was used for vector production assays. In most reports, typical titers range from 104 to 105 adenovirus total particles per cell, whereas IP are 1 log lower.2, 18, 19 Although high viral productivity is obtained using SF medium with adherent monolayer cells in static cultures, the fast decrease of pH after infecting cells in spinner cultures may be responsible for the 20-fold lower productivity. By controlling pH in a 2-l stirred tank bioreactor during process validation, the productivity can be restored to yields comparable with those in static cultures. The optimal harvest times described for adenoviruses are in the range of 40–48 hpi.20, 21 However, under stirring conditions the maximum yield of ΔE1 CAV-2 was attained earlier. Considering that similar cell-specific productivity was obtained in 2-l stirred tank bioreactor (24–36 hpi) and static cultures (40 hpi), we speculate that stirring promotes MDCK cell metabolic activity, which ultimately leads to faster virus production.

CAV-2 vector purification consisted of three main DSP steps (clarification, purification and polishing).

The clarification step consisted of a combination of centrifugation and microfiltration, because a low recovery yield was obtained when only a microfiltration was performed. This is likely due to the entrapment of viral vectors in the pores of membrane or aggregate formation between viruses and cell debris.

Mediated by the strong ion-exchange properties of the hexon protein, anion exchange chromatography is the most suitable purification step of almost all scalable purification process strategies for human adenovirus.20 The monolithic column was a suitable strategy for ΔE1 CAV-2 vector purification. Monolithic column strategy simultaneously reduces the working volume at least 10 times and increases purity, thereby reducing the process time and associated operation costs. The chromatographic profile of CAV-2, when compared with that obtained for human adenovirus type 5 vector, indicated that CAV-2 requires low conductivities to elute (21 mS cm−1). This is consistent with the lower global net charge of CAV-2 (ref. 22) versus human adenovirus type 5, at the same pH.

Although unattractive for scale-up, size exclusion chromatography is one of the most traditional methods for polishing viral samples and buffer exchange to the appropriate final formulation.23 The use of this step was effective for ΔE1 CAV-2 vector recovery yield and separation from low-molecular-weight impurities. However, core bead prototype matrix can overcome the limitations of size exclusion chromatography (sample volume, dilution of sample and operation time). Moreover, the protein removal performance of the core bead prototype permits the increase of sample loads up to 100-fold while maintaining purity. This core bead prototype matrix is a promising alternative for polishing, making them attractive for purification process at larger scales. The overall recovery yields obtained using the best purification strategy for CAV-2 vectors are within the range of recovery yields commonly described for human AdVs (30–80%).24, 25, 26 Still, the amount of DNA quantified in the final sample was 10 times higher than the one described for human adenovirus.26 This amount might surpass the limit of 10 ng of DNA/dose specified by Food and Drug Administration for adenovirus vectors’ clinical use. Thus, additional downstream operations to reduce impurities might be necessary. Increasing the amount of nuclease added in the beginning of downstream process or DNA precipitation can be suitable options to overcome this.20, 27

In conclusion, MDCK-E1 cells were an efficient producer cell line for ΔE1 CAV-2 vectors. Cultivation of MDCK-E1 cells in a stirred tank bioreactor readily yielded a viable cell concentration and a high ΔE1 CAV-2-specific productivity at 36 hpi. For manufacturing purposes, faster cell growth and earlier virus production time point improve process economics. The purification process for ΔE1 CAV-2 vectors can be performed using a simple scalable procedure consisting of three steps and having an anion exchange monolithic column as the key step for capture, concentration and intermediate purification, with a 63% recovery yield of infectious CAV-2 particles. The work described herein presents the first results concerning the stirring cell culture of MDCK-E1 cells for the production of CAV-2 vectors using SF medium and the best purification strategy in a scalable perspective, providing valuable information for bioprocess implementation.

Materials and methods

Cell lines and culture media

DKZeo cells, generated from DK cell line as described by Kremer et al.,6 were maintained in DMEM (Gibco, Paisley, UK) with 10% (v/v) FBS (Gibco) and 1% (v/v) non-essential amino acids (Sigma-Aldrich, St Louis, MO, USA). MDCK-E1 cells were obtained by transfecting MDCK (ECCAC 84121903) with pCI-neo plasmid (Promega, Madison, WI, USA) harboring E1 gene from CAV-2, using a similar approach to the one described for DKZeo cells.2, 28 MDCK-E1 cells were cultivated in MEM (Sigma) with 10% (v/v) FBS, 2 mM glutamine (Gibco) and 1% (v/v) non-essential amino acids (Sigma). These cells were sequentially adapted to Optipro SF medium (Gibco) with 4 mM glutamine (Gibco) by increasing the percentage of Optipro medium in subculture passages. Cells were subcultured twice a week in 150 cm2 t-flasks and maintained in an incubator with a humidified atmosphere of 5% CO2 in air at 37 °C. For splitting or collecting cells, the monolayer was washed with phosphate-buffered saline, incubated with 0.25% trypsin-EDTA (Gibco) at 37 °C until detachment started to become evident (up to 15 min) and cells suspended in culture medium. Cells were seeded directly using culture media containing FBS (MEM and DMEM), or pelleted at 300 g during 10 min and re-suspended in fresh medium when SF medium was used.

Cell growth assays

MDCK-E1 growth assays in static cultures were performed using an inoculum of 1 × 104 cells cm−2 in 25-cm2 t-flasks with a working volume of 10 ml.

The stirred cultures were performed in 125-ml spinner flasks (Wheaton, Millville, NJ, USA) previously siliconized with a solution of dimethyldichlorosilane in toluene (Merck, Darmstadt, Germany) to prevent sticking of cells and microcarriers to the flask’s walls.

Two non-porous microcarriers, Cytodex 3 and Cytodex 1 (GE Healthcare, Uppsala, Sweden), were prepared according to the manufacturer′s instructions. The growth curves were performed using an inoculum of 1 × 105 and 2 × 105 cells ml−1, a microcarrier concentration of 2 and 3 mg ml−1, and a stirring of 50 r.p.m.

Cell concentration determination

Cells were counted using a Fuchs–Rosenthal haemocytometer chamber with the trypan blue exclusion method. In static cultures, cells were counted directly from cell suspension after trypsin detachment. In stirred/microcarrier cultures, a similar procedure was adapted. From a 1-ml sample, microcarriers were let to settle down and supernatant removed (800 μl). Phosphate-buffered saline was added in the same volume to wash the cells and removed after new microcarriers settled down. An equal volume of trypsin was added, cells were incubated at 37 °C until detachment was evident, suspended by pipetting up and down and finally counted.

ΔE1 CAV-2 vector stock preparation

CAVGFP, an ΔE1 CAV-2 vector, was prepared as described previously.2 For stock preparation, 150 cm2 t-flasks with DKZeo cells at a confluency of 80–90% were infected with a MOI of 5 (IP) and medium exchange at time of infection. Forty hour post infection (hpi) cells were collected and lysed with 0.1% (v/v) Triton 100x (Sigma-Aldrich). The lysate was clarified by centrifugation at 3000 g during 10 min at 4 °C and purified by CsCl gradients as described previously.2 The purified vectors were stored in phosphate buffered saline with 10% (v/v) glycerol in aliquots at −85 °C. The IP/physical particles (PP) ratio was 1:21 based on flow cytometry analysis and real-time PCR.

CAVGFP production assays

The infection assays were performed in tissue-culture Petri dishes with 10 cm diameter (BD Biosciences, Franklin Lakes, NJ, USA) and 125 ml spinner vessels (Wheaton). A MOI of 5 (IP) was used and the medium was exchanged at the time of infection. When using tissue-culture petri dishes, 8 × 104 cells cm−2 were seeded and infected 24 h after with a cell confluence of 80–90% and a working volume of 10 ml. In spinner flasks cultures, 1 × 105 cells ml−1 and 2 × 105 cells ml−1 were inoculated using MEM and Optipro media, respectively. Cells were infected at the end of the exponential growth phase (48 h). Extracellular viruses were collected from the cell supernatant. Intracellular viruses were collected by disrupting cells in a known volume of lysing buffer (Tris/HCl 10 mM, pH 8.0 and 0.1% (v/v) Triton-X 100). The resulting intracellular and extracellular samples were clarified at 3000 g for 10 min at 4 °C and stored at −85 °C.

Titration of infectious CAVGFP particles

Quantification of IP was performed by monitoring the expression of GFP reporter gene using slightly modifications to the method described by Ferreira et al.29 Briefly, 4 × 104 MDCK-E1 cells were seeded in 24-well plates and infected with serial dilutions of viral suspensions. Cells were trypsin-detached 24 hpi and the percentage of GFP-positive cells determined by flow cytometry (CyFlow Space, Partec, Münster, Germany).

Measurement of total CAVGFP particles by real-time PCR

CAVGFP DNA was extracted and purified by High Pure Viral Nucleic Acid Kit (Roche Diagnostics, Penzberg, Germany). Previously purified and quantified plasmid coding for the ΔE1 CAV-2 vector genome (pJB19) 30 was used as standard.

Amplifications were performed by the Light Cycler system with ‘Fast Start Master SYBR Green I kit’ (Roche Diagnostics) based on the method developed previously.31 For a 20 μl PCR reaction, 10 μl DNA templates and 10 μl of mastermix were added to each capillary. The mastermix content was 3 mM MgCl2 and 0.5 μM of each primer. The forward primer used was 5′-AAATATACAGGACAAAGAGGTGTGG-3′ and the reverse 5′-GAACTCGCCCTGTCGTAAAA-3′. The standard program was performed according in Light Cycler RT-PCR (Roche Diagnostics) and analyzed using the Light Cycler software. A standard curve was obtained by linear regression analysis of the threshold cycle (Ct) value (y axis) versus the log of the initial copy number present in each sample dilution (x axis). PCR efficiency (E) was calculated as E= 10(−1/slope) – 1 and corresponded to 2.021. The error associated was 0.0849. The PCR limit of detection was 1 × 102 copies per reaction, detected using 40 cycles. The viral DNA loss during DNA extraction/purification showed to be constant and was further considered to calculate PP concentration.31

Protein and DNA analysis

Total protein analyses were carried out using the Pierce BCA kit 23225 (Thermoscientific, Rockford, IL, USA); total DNA analyses were conducted using the PicoGreen dsDNA Assay Kit (Molecular Probes, Invitrogen). Absorbance was read at 562 nm to determine total protein, and green fluorescence was measured in a fluorimeter for detecting DNA. The PicoGreen assay had a limit detection of 25 pg ml−1 of dsDNA (50 pg dsDNA in a 2 ml assay volume).

Protein profile of CAVGFP by SDS-PAGE and western blot

Viral particles were precipitated in absolute ethanol at −20 °C overnight, centrifuged at 10 000 g for 10 min and concentrated in the SDS-PAGE loading-buffer at 30 μg of protein per total volume to apply on gel. SDS-PAGE was performed according to NuPAGE Novex Bis-Tris Mini Gels protocol using the Xcell SureLock equipment (Invitrogen). Western blot analysis was carried out using a polyclonal rabbit anti-CAV-2 primary antibody.32 Proteins were revealed via BCIP/NBT substrate (Pierce) after incubation with an alkaline phosphatase-conjugated anti-rabbit secondary antibody (Sigma-Aldrich).

Size and electrodynamics measurements of CAVGFP

Size and zeta potential of viral particles were determined by dynamic light scatering using a Zetasizer Nano-ZS Series ZEN3600 equipped with a 633 nm He-Ne laser (Malvern, Worcestershire, UK). Zeta potential was measured using a 40 μl sample in 660 μl 10 mM phosphate buffer (Sigma-Aldrich) for pH ranging from 3 to 10 in 10 × 10 mm plastic cuvette (Sarstedt, USA) using a dip-cell (Malvern). Particles size was determined using the same sample and buffer volumes at pH 8.

Bioreactor culture and CAVGFP production

MDCK-E1 cells were cultivated in siliconized 2-l stirred tank bioreactors (Biostat Q-Plus, Sartorius Stedim Biotech) equipped with three blade impellers, under defined conditions (working volume: 1.5 l; pH: 7.4; temperature: 37 °C; surface aeration at 40% pO2; agitation rate: 50–60 r.p.m.). Cells (2 × 105 cells ml−1) were seeded on 3 mg ml−1 of Cytodex 1 microcarriers placed inside glass bottles and transferred to bioreactor vessels. Culture medium Optipro was supplemented with 5% (v/v) FBS during inoculation (12 h). Medium replacement was done by stopping agitation and letting the microcarriers settle down. After 2 days cultivation (1 × 106 cells ml−1), the medium was exchanged to regular Optipro (without FBS) and cells were infected with MOI 5.

Harvest and clarification

Harvesting was performed at the end of bioreaction (approx. at 42 hpi). To promote cell lysis, 0.1% Triton X-100 (Merck) was added to the bulk and incubated for 30 min at 37 °C with Benzonase (50 U ml−1). After this, microcarriers were left to settle and the resulting supernatant was used for further processing. Clarification was first performed by microfiltration using Sartobrand P 0.45 μm filter capsule (Sartorius Stedim Biotech) at 70 ml min−1.

Centrifugation was performed at 10 000 g, 15 min at 4 °C. A Sartopure 2 membrane capsule made of polyethersulfone (Sartorius Stedim Biotech) with a prefilter of 0.8 μm+0.45 μm filter retention and 50 cm2 of filtration area was used.


ÄKTAexplorer 100 and 10 systems (GE Healthcare) were used to perform liquid chromatography purifications; these systems were equipped with an UV, conductivity and pH meters and were controlled on-line with the coupled UNICORN software (GE Healthcare). Buffer Tris-HCl (10 mM, pH 8.0) supplemented with 2 mM of MgCl2 buffer was used for all the columns’ equilibration.

Anion exchange chromatography

Two strong anion exchange chromatography devices, CIM QA-8f Tube Monolithic Column (containing quaternary amine ligands) and Sartobind Q 75, were used. Tris-HCl (10 mM, pH 8.0) supplemented with 2 mM of MgCl2 plus 1 M of NaCl was used as mixing buffer for the elution steps.

The monolithic column was studied as a capture step for CAV-2 purification. Four flow rates were evaluated: 5, 10, 20 and 40 ml min−1; eight column volume linear salt elution dients were performed; isocratic salt elution steps were tested using 75, 200, 300 and 500 mM NaCl steps at 40 ml min−1.

Size exclusion chromatography (gel filtration)

Prepacked Superdex 200 10/300 GL (GE Healthcare) with 1 cm diameter and 30 cm length was used for this study. It has an optimal resolution range from 5 to 200 kDa of separation. Buffer Tris-HCl (10 mM, pH 8.0) supplemented with 2 mM of MgCl2 was used at 1 ml min−1 (76 cm h−1).

Core bead prototype chromatography

A core bead prototype matrix (generously provided by GE Healthcare) was used for the polishing step study. The beads contain functional groups (ligands) strictly inside the spherical matrix enabling binding of fragmented host cell DNA and host cell protein impurities by ion-exchange and hydrophobic interaction while letting the virus particles flow freely in the inter-bead space. One prepacked Hiscreen column mixing different types of liquid chromatography was used. The column was equilibrated with 10 mM Tris-HCl pH 8.0 supplemented with 2 mM of MgCl2 and the sample was loaded at 10 mS cm−1 and 150 or 300 cm h−1.


  1. 1.

    , , . From the first to the third generation adenoviral vector: what parameters are governing the production yield? Biotechnol Adv 2009; 27: 133–144.

  2. 2.

    , , , . Canine adenovirus vectors: an alternative for adenovirus-mediated gene transfer. J Virol 2000; 74: 505–512.

  3. 3.

    , . Frequency, proliferation, and activation of human memory T cells induced by a nonhuman adenovirus. J Virol 2005; 79: 14595–14605.

  4. 4.

    , , . Characterization of cis-acting sequences involved in canine adenovirus packaging. Mol Ther 2001; 3: 631–640.

  5. 5.

    , , . An update on canine adenovirus type 2 and its vectors. Viruses 2010; 2: 2134–2153.

  6. 6.

    , , , . Preferential transduction of neurons by canine adenovirus vectors and their efficient retrograde transport in vivo. Faseb J 2001; 15: 2283–2285.

  7. 7.

    , , , , , . Serum free cell culture: the free access online database. Toxicol In Vitro 2006; 20: 395–400.

  8. 8.

    , , , , , . Production of first generation adenoviral vectors for preclinical protocols: amplification, purification and functional titration. J Biosci Bioeng 2011; 112: 415–421.

  9. 9.

    , , . HPLC purification of adenoviral vectors. Methods Mol Biol 2010; 594: 395–408.

  10. 10.

    , , , . Purification of adenoviral vectors by combined anion exchange and gel filtration chromatography. J Gene Med 2009; 11: 978–989.

  11. 11.

    , , , , , et al. Concentration and purification of rubella virus using monolithic chromatographic support. J Chromatogr B Analyt Technol Biomed Life Sci 2011; 879: 981–986.

  12. 12.

    , , , , , et al. Reduced surface area chromatography for flow-through purification of viruses and virus like particles. J Chromatogr A 2011; 1218: 3973–3981.

  13. 13.

    , , . Serum-free influenza virus production avoiding washing steps and medium exchange in large-scale microcarrier culture. Vaccine 2006; 24: 3261–3272.

  14. 14.

    , , , , . Comparison of large-scale mammalian cell culture systems with egg culture for the production of influenza virus A vaccine strains. Vaccine 2001; 19: 3444–3450.

  15. 15.

    , , , , . Metabolism of MDCK cells during cell growth and influenza virus production in large-scale microcarrier culture. Vaccine 2004; 22: 2202–2208.

  16. 16.

    , , , , . Substitution of glutamine by pyruvate to reduce ammonia formation and growth inhibition of mammalian cells. Biotechnol Prog 2005; 21: 58–69.

  17. 17.

    , . Trafficking and propagation of canine adenovirus vectors lacking a known integrin-interacting motif. Hum Gene Ther 2001; 12: 1815–1823.

  18. 18.

    , . Production of adenovirus vector for gene therapy. Biotechnol Adv 2003; 20: 475–489.

  19. 19.

    , , , . Effect of re-feed strategies and non-ammoniagenic medium on adenovirus production at high cell densities. J Biotechnol 2005; 119: 272–280.

  20. 20.

    , , , , , . Production and formulation of adenovirus vectors. Adv Biochem Eng Biotechnol 2005; 99: 193–260.

  21. 21.

    , . Development and optimization of an adenovirus production process. J Gene Med 2004; 6: 184–192.

  22. 22.

    , , , , , et al. Three-dimensional structure of canine adenovirus serotype 2 capsid. J Virol 2008; 82: 3192–3203.

  23. 23.

    , , . Overview of current scalable methods for purification of viral vectors. Methods Mol Biol 2011; 737: 89–116.

  24. 24.

    , , , , , et al. Adenovirus vector production and purification. Curr Gene Ther 2010; 10: 437–455.

  25. 25.

    . Good manufacturing practice production of adenoviral vectors for clinical trials. Hum Gene Ther 2005; 16: 281–291.

  26. 26.

    , , , , . Towards purification of adenoviral vectors based on membrane technology. Biotechnol Prog 2008; 24: 1290–1296.

  27. 27.

    , , , , . Development of a novel adenovirus purification process utilizing selective precipitation of cellular DNA. Biotechnol Bioeng 2005; 91: 12–21.

  28. 28.

    , , , , , et al. A recombinant E1-deleted canine adenoviral vector capable of transduction and expression of a transgene in human-derived cells and in vivo. Human Gene Ther 1997; 8: 2103–2115.

  29. 29.

    , , , , , et al. 293 cell cycle synchronisation adenovirus vector production. Biotechnol Prog 2009; 25: 235–243.

  30. 30.

    , , . Long-term in vivo transduction of neurons throughout the rat CNS using novel helper-dependent CAV-2 vectors. FASEB J 2004; 18: 391–393.

  31. 31.

    , , , , , . A real-time PCR assay for quantification of canine adenoviral vectors. J Virol Methods 2010; 163: 129–136.

  32. 32.

    , , , , , et al. CAR-associated vesicular transport of an adenovirus in motor neuron axons. PLoS pathogens 2009; 5: e1000442.

Download references


This work was supported by the Fundação para a Ciência e Tecnologia (FCT)—Portugal (through the projects PTDC/BIO/69452/2006, PTDC/EBB-BIO/119501/2010 and PTDC/EBB-BIO/118615/2010) and the European Commission (BrainCAV HEALTH – HS_2008_222992). Paulo Fernandes acknowledges the FCT for his PhD grant (SFRH/BD/70810/2010). We acknowledge Eng. Marcos Sousa for the technical support in bioreaction. We also acknowledge the Sartorius Stedim Biotech for providing the membranes, the BIA Separations for providing the monolithic columns, the GE Healthcare for providing the core bead prototype matrix and for the technical advice.

Author information


  1. Animal Cell Technology Unit, Instituto de Tecnologia Química e Biológica, Universidade Nova de Lisboa (ITQB-UNL), Oeiras, Portugal

    • P Fernandes
    • , A S Coroadinha
    •  & P M Alves
  2. Instituto de Biologia Experimental e Tecnológica (IBET), Oeiras, Portugal

    • P Fernandes
    • , C Peixoto
    • , V M Santiago
    • , A S Coroadinha
    •  & P M Alves
  3. Institut de Génétique Moléculaire de Montpellier, CNRS—Universities of Montpellier I and II, Montpellier, France

    • E J Kremer


  1. Search for P Fernandes in:

  2. Search for C Peixoto in:

  3. Search for V M Santiago in:

  4. Search for E J Kremer in:

  5. Search for A S Coroadinha in:

  6. Search for P M Alves in:

Competing interests

The authors declare no conflict of interest.

Corresponding author

Correspondence to P M Alves.

About this article

Publication history







Further reading