Pilot-scale process development for recombinant adeno-associated virus (rAAV) production based on high-density Sf9 cell culture

Abstract

Background

In recent years, gene therapy drugs have been widely marketed, and their effectiveness and potential have been confirmed. Thus, increasing their production on an industrial scale is critical. Recombinant adeno-associated viruses (rAAVs) are optimal vectors for gene therapy applications, and the baculovirus expression vector system (BEVS), which is based on Sf9 cell culture, is a common tool for rAAV production.

Methods

In this work, an Sf9 cell fed-batch process was developed using shake flasks. In the laboratory-scale bioreactor, four processes were selected as the key factors when carrying out the orthogonal experiment. On the basis of the equal P/V principle and considering the problem posed by air bubbles, a pilot-scale level bioreactor process was established.

Results

Here, we describe a method in which a BEVS was used to produce rAAV vectors, with the cell density increasing to 22.8 × 10⁶ cells/mL and the rAAV titre increasing to 20 × 10¹¹ VG/mL upon adding feed material. By resolving the problems associated with high-density cell culture and air bubbles, this process was successfully scaled to a 50 L pilot-scale level.

Conclusions

This successful experiment not only provides a technological basis for further scale-up but also guarantees product capacity. We hope that this development process can provide reference data for studying cell culture-based drug production.

Introduction

Gene therapy involves the introduction of exogenous genetic material into patients to alter the expression of genes or proteins and achieve therapeutic effects, providing the possibility of a one-dose cure for many currently incurable genetic diseases [1, 2]. Among gene therapy tools, adeno-associated virus (AAV) vectors have certain advantages that have facilitated the production of ideal therapeutic effects in the preclinical and clinical treatment of various human genetic diseases; these advantages include their nonintegration into the target cell genome, low immunogenicity, high delivery efficiency, and wide range of target cell types [3, 4]. Two main AAV production systems are used in industry: transient transfection of human HEK293 cells and baculovirus infection of Spodoptera frugiperda (Sf9) insect cells [4, 5].

Recently, 8 rAAV drugs have been launched worldwide for use in the treatment of haemophilia and ophthalmological, muscle, and nerve diseases. The cell matrices for these drugs are produced by using the above two cell lines. Compared with plasmid transfection, the baculovirus expression vector system (BEVS)-based method has the advantages of high virus yield, excellent scalability and relatively low cost, and the it has the ability to more easily meet the large demand for rAAV vectors in clinical practice [6, 7]. At present, batch production is the most commonly used method for generating rAAVs in Sf9 cell culture, with a maximum cell density of 7.5 × 10⁶ cells/mL and a maximum rAAV titre of 6 × 10¹¹ VG/mL [8–11]. Fed-batch methods are less commonly used, and these methods have a maximum cell density of 12 × 10⁶ cells/mL and a maximum rAAV titre of approximately 3 × 10¹¹ cells/mL [8, 10].

Scaling up small-scale production processes to pilot- and large-scale commercial production is one of the key steps in drug process development. At present, large-scale cell cultures are performed mainly with suspensions, and the most commonly used vessels are stirred reactors. Sf9 insect cells are sensitive to the shear forces generated by the stirring paddles, and this sensitivity increases after baculovirus infection. Therefore, developing a suitable scale-up process for Sf9 culture and rAAV packaging is very important [12, 13]. The two most commonly used approaches when scaling up bioreactor processes are consistent in the volumetric mass transfer coefficient (kla) and the impeller power input per unit volume (P/V); other scale-up methodologies include consistent in the Reynolds number (Re), impeller tip speed, and mixing time [14–16]. In current industrial production, the maximum scale for the Sf9/baculovirus system is 200 L, the highest viable cell density is approximately 7 × 10⁶ cells/mL, and the highest rAAV titre is approximately 2 × 10¹¹ VG/mL [13, 17, 18].

In this work, an Sf9 cell fed-batch process was developed in shake flasks. The maximum cell density was 28.6 × 10⁶ cells/mL, which was 119% greater than that of the batch process. rAAV was produced on the basis of this cell culture process; the maximum cell density reached 22.8 × 10⁶ cells/mL, and the viral vector titre was 16 × 10¹¹ VG/mL. On the basis of this shake flask process, small-scale and pilot-scale reactor processes were developed. In the 1 L bioreactor, we focused on the viral stage and selected four reactor parameters for the L9(3⁴) orthogonal experiment. The rAAV titre was increased to 24.1 × 10¹¹ VG/mL after optimization.

On the basis of the equal P/V principle and considering the problem posed by bubbles, a pilot-scale reactor process was established. The results revealed that the cell density, gas partial pressure, and metabolite profiles were similar at different scales and that there was no significant difference in the rAAV titre. This paper describes a pilot-scale process for the production of rAAVs via high-density insect cell culture. In this process, the cell density and rAAV titre were increased by feeding, and the problem posed by air bubbles in high-density culture was solved by adjusting the stirring speed and temperature. This process provides a reference for further scale-up methods and can be expanded to the production of other rAAV products and insect cell culture-based drugs.

Materials and methods

Cell culture and rAAV vector production

Sf9 cells (Gibco, cat. no. 12659017) were cultivated in Sf900 III (Gibco, cat. no. 12658019). The shake flasks were incubated in a shaking incubator (Radobio, 50 mm shaking diameter) at 27 °C and 130 rpm. Glass bioreactors (1 L) (Eppendorf AG, Hamburg, Germany) and a single-use bioreactor (50 L) (Pall, Berliner, Germany) were used in this study. Sodium bicarbonate (10% w/v) and CO₂ gas were used to adjust the pH. The dissolved oxygen (DO) was maintained through sparging of air and oxygen gas mixture.

Sf9 cells were inoculated at a viable density (5 × 10⁵ cells/mL). InsectPro (BasalMedia, cat. no. H861KJ) was added to the media as a supplement. Sf9 cells were infected for 168 h with rBV-GOI (Factor IX) and rBV-CapA1 (serotype 6) from baculovirus banks, each with an MOI of 0.05 IVP/cell. The viable cell concentration and cell viability were determined via an IC1000 instrument (Countstar, China) using the trypan blue dye exclusion method [19]. The concentrations of glucose, glutamine, lactate, ammonia and lactate dehydrogenase (LDH) in the cell suspension were measured daily with a Cedex bioanalyzer (Roche, Switzerland). Cultures were terminated when the cell viability decreased to less than 30%.

Recombinant baculovirus (rBV) construction

Two rBVs were produced by transfecting Sf9 with bacmids containing certain GOIs or Rep/Cap genes to yield rBV-GOI or rBV-CapA1, respectively, according to the instructions for the Bac-to-Bac Baculovirus Expression System. Flow cytometry (ACEA Biosciences, USA) was used to quantify the expression of the labelled BV-based viral surface protein gp64 [20].

rAAV vector harvesting

The cells and supernatant were collected by centrifugation (12,000 rpm, 10 min). The cells were lysed with lysis buffer containing 1× phosphate-buffered saline (PBS), 1% Triton X-100 and 10 mmol/L MgCl₂. After 1 h, the mixture was centrifuged at 12,000 rpm for 10 min to remove cell debris. Then, the cell culture supernatant and lysis supernatant were combined as the rAAV supernatant, to which BenzoNuclease was added at a working concentration of 100 U/mL (Novoprotein, cat. no. GMP-1707) for 1 h of incubation at room temperature.

rAAV vector quantification

The rAAV titre was quantified via quantitative polymerase chain reaction (qPCR) using CMV-specific primers as described previously [21].

Statistical analysis

The data in this work were analysed with GraphPad Prism 8 and are presented as the means ± standard deviations (SDs). One-way ANOVA was used for comparisons. A P value < 0.05 was considered to indicate statistical significance.

Results

Comparison of the batch and fed-batch processes

In our laboratory, we developed a fed-batch process for Sf9 cells. When 2% of the total volume of feed material was added each day starting on the fifth day, the maximum cell density increased by 119% compared with that in batch culture, reaching 28.6 × 10⁶ cells/mL (Fig. 1). A rAAV was produced on the basis of this cell culture process. Compared with that in batch culture, in the fed-batch process, the maximum cell density increased by 85% to 22.8 × 10⁶ cells/mL (Fig. 1), and the rAAV titre increased approximately 9-fold to 16 × 10¹¹ VG/mL (Fig. 2).

Fig. 1.

Comparison of cell growth profiles between different cell lines and processes (n = 3); mean ± standard deviation

Fig. 2.

Effects of different processes on the rAAV titre; mean ± standard deviation

Optimization of the laboratory-scale bioreactor process

The bioreactor process parameters after BV infection were determined on the basis of the rAAV production process carried out in shake flasks. Here, we refer to pre-BV infection as the cell stage and postinfection as the viral stage. On the basis of previous study results, the process parameters for the cell stage in the reactor were as follows: temperature: 27.5 °C, pH: 6.0, DO: 40, and agitation speed: 300 rpm. For the virus stage, we selected four reactor parameters to conduct the L9(3⁴) orthogonal experiment. Tables 1 and 2 present the detailed experimental design, and the results are shown in Table 2. Compared with a temperature of 27.5 °C during the cell culture period, lowering the temperature to 25 °C negatively affected the rAAV titre, which decreased by 32%. The viral titre at 30 °C was similar to that at 27.5 °C. Compared with that in the cell stage at pH 6.0, the rAAV titre was greater at a more alkaline pH (6.3), whereas an acidic pH of 5.7 had the opposite effect. The DO had no significant effect on the viral titre at the three levels tested. We therefore set the DO for subsequent processes at 40, which was the same as that used for the cell stage. The optimal stirring speed in the virus stage was 250 rpm, which was slightly lower than that in the cell culture stage. Finally, the optimal conditions were combined for rAAV packaging validation. The obtained rAAV titre was 24.1 × 10¹¹ VG/mL, which was greater than the highest titre of 17.6 × 10¹¹ VG/mL obtained in the orthogonal experiment.

Table 1.

A subset of the operating parameters investigated during the orthogonal experiment and their associated ranges

	T(A)	pH(B)	DO(C)	Agit(D)
1	25	5.7	20	250
2	27.5	6.0	40	300
3	30	6.3	60	350

Table 2.

Design of the L9(3⁴) orthogonal experiment and the results. (T1, T2, and T3 are the mean values of each factor at the same level)

	T(A)	pH(B)	DO(C)	Agit(D)	Titer(×1011 VG/ml)
1	1	1	1	1	8.4
2	1	2	2	2	11.1
3	1	3	3	3	7.6
4	2	1	2	3	6.8
5	2	2	3	1	17.6
6	2	3	1	2	15.3
7	3	1	3	2	8.6
8	3	2	1	3	9.6
9	3	3	2	1	19.7
T1	9.03	7.93	11.10	15.23
T2	13.23	12.77	12.53	11.67
T3	12.63	14.20	11.27	8.00

Scaling up to 50 L and verification

To meet production needs, we scaled up the 1 L rAAV production process to 50 L. Table 3 lists the parameters of the two reactors. The stirring speed in the 50 L reactor was 92 rpm for the cell stage on the basis of the equal P/V principle; the stirring speed was reduced to 77 rpm for the virus stage on the basis of this same principle. However, during the actual production process, many bubbles appeared in the virus stage. Moreover, as many of these bubbles burst, the cells became obviously damaged, which led to a decrease in the rAAV titre (4.3 × 10¹¹ VG/mL). Therefore, we optimized the stirring speed (87 rpm) and culture temperature (30 °C) in the 50 L reactor for the virus stage. The stirring speed and fluid parameters of the two reactors are shown in Table 4. During validation, the other process parameters were maintained. These results are shown in Figs. 3 and 4. Notably, the obtained cell density curves were similar, with a maximum density of approximately 20 × 10⁶ cells/mL. Additionally, the rAAV titre at the 50 L scale was 21.3 × 10¹¹ VG/mL, which was slightly lower than that at the 1 L scale, but the difference was not significant. Thus, these data indicate that this process was effectively scaled up to the pilot scale.

Table 3.

Geometric parameters of the 1 L and 50 L fermenters

	1-L fermenter	50-L fermenter
Diameter (T, mm)	120	380
Bottom Type	Dished	Dished
Filled volume (VL, L)	1	50
Power Number (Po)	1.3	1.9
Impellers (Di, mm)	45	185

Table 4.

The agitation speed and fluid parameters for the 1-L and 50-L reactor

	1-L cell stage	50-L cell stage	1-L virus stage	50-L virus stage
Stirring speed (rpm)	300	92	250	87
P/V (w/m3)	29.99	29.69	17.35	25.10
tip speed (m/s)	0.71	0.89	0.59	0.84
shear stress (N/m2)	0.58	0.49	0.44	0.45
eddy size (µm)	41.42	45.08	47.48	47.01

Fig. 3.

Comparison of viable cell density and viability from bioreactors on different scales (n = 3); mean ± standard deviation

Fig. 4.

Effects of the bioreactor scale on the rAAV titre (n = 3); mean ± standard deviation

Comparison of pO2, pCO2 and metabolism at different scales

We analysed the pO₂, pCO₂ and metabolism at different scales daily. As shown in Fig. 5, from days 0 ∼ 5, the pO₂ gradually decreased to approximately 50 and then fluctuated but remained at approximately 50; notably, this pO₂ levels can satisfy the needs of the cells. Additionally, the pCO₂ gradually increased from days 0 ∼ 5 but did not increase further thereafter; the overall value remained below 40, which did not significantly impact the cell cultures. Figure 6 shows that glucose (with a starting concentration of 10 g/L) was continuously consumed and almost completely consumed by the end of the culture period. The accumulation of lactic acid was slow, increasing from day 11 to the end of culture on day 14 (1.2 g/L), which did not negatively affect the cells. LDH is produced mainly by lysed cells in media, and the LDH can be used to determine the total number of lysed cells. Cell damage caused by air bubbles can lead to an increase in the LDH. As mentioned in the previous section, to address the negative effect of air bubbles, the stirring speed in the 50 L reactor during the virus stage was increased from 77 rpm to 87 rpm, and the culture temperature was increased to 30 °C. These changes can increase gas‒liquid transfer and decrease the number of air bubbles, ultimately reducing the degree of damage caused by air bubbles. The maximum amount of LDH in the 50 L reactor was only 200 units/L, which was significantly lower than the 600 units/L in the 1 L reactor, indicating that the negative effect of the air bubbles was effectively resolved during the scale-up process. As shown in Fig. 7, glutamine and ammonia were continuously consumed starting at the beginning of culture and stabilized at approximately 1 mmol/L until the sixth day, after which consumption rebounded at the end of culture. In general, the trends of the gas and metabolism curves were consistent at the different scales, and the values were similar, indicating that the cell culture processes in the two reactors were similar and that the scale-up linearity was good.

Fig. 5.

Comparison of pO₂ and pCO₂ between bioreactors at different scales (n = 3); mean ± standard deviation

Fig. 6.

Comparison of glucose, lactic acid and LDH levels between bioreactors at different scales (n = 3); mean ± standard deviation

Fig. 7.

Comparison of glutamine and ammonia levels between bioreactors at different scales (n = 3); mean ± standard deviation

Discussion

AAV vectors are the main gene delivery platform used in modern gene therapy applications. Thus, it is critical to produce rAAV drug products on a large scale to promote clinical trials and commercial manufacturing. Currently, two cell lines are used for rAAV production: HEK293 and Sf9 [22]. The baculovirus-insect cell culture system has the advantages of high product titre, simple processing, and low cost [23, 24]. Currently, a majority of insect cell-based rAAV production processes involve batch cultures, with fewer fed-batch processes described [25–28]. Our laboratory previously developed a fed-batch process in which the maximum cell density doubled to 28.6 × 10⁶ cells/mL compared with that in batch culture (Fig. 1). Compared with that in batch culture, the maximum cell density in the fed-batch process increased by 85% to 22.8 × 10⁶ cells/mL, and the rAAV titre increased approximately 9-fold to 16 × 10¹¹ VG/mL (Fig. 2). On this basis, we attempted to scale up this rAAV production process to the pilot scale.

We believe that the microenvironmental requirements differ between the cell culture stage and the viral packaging stage in laboratory-scale reactors; in the former, cell growth should be considered, whereas the needs of both viruses and cells should be considered in the latter. In this study, we focused on optimizing the conditions during the virus stage in a laboratory-scale reactor. Culture temperature is one of the key factors in rAAV production and affects the proliferation and viability of cells and viruses. At low temperatures, cells and viruses can retain greater viability, although their metabolism and proliferation rates are reduced. In contrast, at higher temperatures, the proliferation rate increases, but the rate of inactivation also increases. Notably, maintaining viability by decreasing the temperature prolongs the culture time, and the viability of cells and viruses gradually decreases with increasing culture time [29, 30]. In this study, a temperature of 25 °C had an adverse effect on the rAAV titre, but temperatures of 27.5 ∼ 30 °C had no significant effect on the viral vector titre. Considering the stability of the product, a relatively low temperature of 27.5 °C was chosen. pH affects the microenvironment of cells and viruses. Some studies have reported that alkaline conditions are more conducive to the amplification of viruses, unlike cell cultures [31, 32]. The experimental results were consistent with conclusions described in the literature, and compared with the pH of the cell stage (pH 6.0), a more alkaline environment (pH 6.3) was conducive to viral vector packaging, and higher product titres were obtained. Additionally, the DO had little effect on the viral titre in the range of 20 ∼ 60; thus, the pO₂ was maintained between 40 and 120 mmHg, which was satisfactory for cell growth [33]. The optimal stirring speed (250 rpm) for viral vector packaging was lower than that for the cell stage (300 rpm). We found that a higher stirring speed increased the likelihood of virus inactivation, and virus-infected cells were more sensitive to shear forces. During the virus stage, the rAAV titre could be increased by appropriately reducing the stirring speed. The optimal conditions were validated, and the rAAV titre further increased to 24.1 × 10¹¹ VG/mL under these conditions.

By utilizing the laboratory-scale test technique, we scaled up the process to the pilot scale. The stirring speed during the cell stage was set to 92 rpm on the basis of the equal P/V principle. During the viral stage, we found that, on the basis of this same principle, the optimal stirring speed was 77 rpm. Under these conditions, the ventilation volume was relatively large (> 5 L/min), and bubbles accumulated on the gas‒liquid surface. Studies have shown that cells attached to bubble surfaces at the air–liquid interface are damaged when the bubble bursts. At the 50 L scale, the presence of too many air bubbles reduced both cell viability and density. Moreover, LDH accumulation was greater than 1,000 units/mL. This poor cell culture process eventually led to a decrease in the rAAV titre (4.3 × 10 ¹¹ VG/mL). Increasing the stirring speed can break up the bubbles more thoroughly, increase the gas‒liquid exchange area, and prolong the duration for which the bubbles remain in the medium, all of which have a positive effect on gas dissolution and reduce bubble accumulation. On the basis of experimental results, we increased the stirring speed from 77 rpm to 87 rpm. The stirring speed and fluid parameters are shown in Table 4; each parameter was within the appropriate range [34]. In addition, we adjusted the temperature of the 50 L bioreactor in the virus stage. Table 2 shows that at 30 °C, rAAV production did not significantly decrease. Increasing the temperature can increase the speed of gas dissolution. Therefore, 30 °C was used during the virus stage in the 50 L bioreactor, while the other process conditions remained the same as those at the laboratory scale. The cell culture profiles are shown in Fig. 3. The profiles of cell density were similar on both scales, with a maximum density of approximately 20 × 10⁶ cells/mL. In terms of viability, the cell survival rate began to decline on the 10th day (the third day after virus infection), decreasing to approximately 30%, and the culture was terminated on the 14th day. The viral titres tended to increase after BV infection and peaked168 hours after BV infection. The rAAV titre on the 50 L scale was slightly lower than that on the 1 L scale. This may have been caused by the increased stirring speed during the virus stage on the 50 L scale, which decreased the number of air bubbles. Nevertheless, the rAAV titre at the 50 L scale remained above 20 × 10¹¹ VG/mL and was not significantly different from that at the 1 L scale. The empty rAAV capsid ratio in the 50 L bioreactor process was approximately 52% in the cell fermentation broth and 23% after purification; these empty capsid ratios are acceptable for rAAV production.

During scale-up validation, we detected the pO₂, pCO₂ and metabolites. Oxygen is a key factor in the cell culture process. If the oxygen supply is insufficient, irreversible damage to the cells can occur. While controlling the DO in the reactor, we also sampled the medium daily and measured the pO₂ in the solution offline. Figure 5 shows that the curves of pO₂ were similar at both scales. During the first 6 days, as the cell density increased, the pO₂ decreased and then fluctuated, remaining at approximately 50 mmHg, which is sufficient to meet the needs of the cells. The pCO₂ often increases excessively during high-density mammalian cell culture. However, in this study, the highest density of the insect cells was 20 × 10⁶ cells/mL, and the increase in the pCO₂ was slow, with the value remaining below 40 mmHg. On the basis of prior experience, a pCO₂ below 80 mmHg will not harm cells. Owing to metabolic differences, insect cells have significantly lower CO₂ outputs than do mammalian cells. Thus, during large-scale high-density insect cell culture, CO₂ accumulation is unlikely to occur.

In terms of glucose metabolism, from day 0 to the end of culture, glucose was gradually consumed, with the level decreasing from 10 g/L to 0.5 g/L, and the needs of the cells for a carbon source were met throughout the process. As a byproduct of glucose metabolism, lactic acid is generally considered to have a negative effect on cells when its content exceeds 3 g/L. When lactic acid was detected, we found that fewer lactic acid insect cells secreted. There was almost no increase in the lactic acid content in the first 11 days, and the concentration of lactic acid accumulated by the end of the culture period was only 1.2 g/L. This finding shows that the problem of lactic acid accumulation also does not easily occur during insect cell culture. LDH is present inside cells and is released into the culture supernatant upon cell lysis. Because LDH is stable, it can be used to indicate total cell lysis and thus to understand the quality of the entire cell culture process. Figure 6 shows that in the virus stage (from the seventh day on) at the 1 L scale, owing to the reduction in rotation speed, the number of air bubbles increased, and the accumulation speed of LDH increased significantly compared with that during the previous 7 days. However, there was no decrease in the rAAV titre, which was greater than 20 × 10⁶ VG/mL (Fig. 4). An LDH level of 600 units/L is considered acceptable. At the 50 L scale, because the stirring speed decreased slightly during the viral stage (from 92 rpm to 87 rpm), the rapid increase in LDH observed in the 1 L reactor did not occur. To examine nitrogen metabolism, we measured glutamine and ammonia levels. Glutamine was continuously consumed starting at the beginning of culture, and the concentration plateaued at approximately 1 mmol/L, indicating that the glutamine needs of the cells were met. Figure 7 shows that ammonia was a consumable nitrogen source in this study. The curves for ammonia were quite different from those in the mammalian system, similar to those for microorganisms. After the gases and metabolites were measured at different scales, strong similarities were found, indicating that the scale-up process was linear. This experience provides a theoretical basis for subsequent scale-up production.

Conclusions

Our laboratory developed a fed-batch process for culturing Sf9 cells, with the cell density reaching 28.6 × 10⁶ cells/mL. This process was applied for rAAV production with a cell density of up to 22.8 × 10⁶ cells/mL and a viral titre of 16 × 10⁶ VG/mL. On this basis, we proceeded with process development with a bioreactor and pilot scale-up. In the laboratory-scale bioreactor, we focused on process development during the virus stage. The temperature, pH, DO and stirring speed were selected as the key factors when L9(3⁴) orthogonal experiments were carried out. On the basis of the experimental results, the bioreactor parameters at the virus stage differed from those at the cell stage, and the rAAV titre increased to 24.1 × 10¹¹ VG/mL. On the basis of the equal P/V principle and considering the problem posed by air bubbles, a pilot-scale level bioreactor process was established. The results showed that among the cells at different scales, the trends in the gas partial pressures and the metabolite curves were similar, as were the values, indicating that the cell culture process was scaled up well. The rAAV titres did not significantly differ between the two scales. These results confirmed that this scale-up strategy was consistent with the scale-up pattern for bioreactor production processes. A pilot-scale process was successfully established for the production of rAAVs via high-density Sf9 cell culture. Overall, we believe that this practical method will have a beneficial effect on the costs of AAV products. Importantly, the method presented here is not specific to the case study shown and can be extended to other biotechnological processes.

Author contributions

Author X. provided working concept or design, collected data, drafted the paper, maked important revisions to the paper, and approval of the final version of the paper for publication. Author J. wrote the background of manuscript. Author Y. supervised the project, provided strategic direction, and conducted a thorough review and final approval of the manuscript prior to submission. Author H. supervised the project, provided fund.

Funding

This work was supported by the Zhejiang Key R&D Program (No. 2021C03085).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Authors declare that the results/data/figures in this manuscript have not been published elsewhere, nor are they under consideration by another publisher, and consent for publication upon acceptance by the journal.

Competing interests

The authors declare no competing interests.