SweetBac: A New Approach for the Production of Mammalianised Glycoproteins in Insect Cells

Recombinant production of therapeutically active proteins has become a central focus of contemporary life science research. These proteins are often produced in mammalian cells, in order to obtain products with post-translational modifications similar to their natural counterparts. However, in cases where a fast and flexible system for recombinant production of proteins is needed, the use of mammalian cells is limited. The baculoviral insect cell system has proven to be a powerful alternative for the expression of a wide range of recombinant proteins in short time frames. The major drawback of baculoviral systems lies in the inability to perform mammalian-like glycosylation required for the production of therapeutic glycoproteins. In this study we integrated sequences encoding Caenorhabditis elegans N-acetylglucosaminyltransferase II and bovine β1,4-galactosyltransferase I into the backbone of a baculovirus genome. The thereby generated SweetBac virus was subsequently used for the production of the human HIV anti-gp41 antibody 3D6 by integrating heavy and light chain open reading frames into the SweetBac genome. The parallel expression of target genes and glycosyltransferases reduced the yield of secreted antibody. However, the overall expression rate, especially in the recently established Tnao38 cell line, was comparable to that of transient expression in mammalian cells. In order to evaluate the ability of SweetBac to generate mammalian-like N-glycan structures on 3D6 antibody, we performed SDS-PAGE and tested for the presence of terminal galactose using Riccinus communis agglutinin I. The mammalianised variants of 3D6 showed highly specific binding to the lectin, indicating proper functionality. To confirm these results, PNGase A released N-glycans were analyzed by MALDI-TOF-MS and shown to contain structures with mainly one or two terminal galactose residues. Since the presence of specific N-glycans has an impact on antibodies ability to exert different effector functions, we tested the binding to human Fc gamma receptor I present on U937 cells.


Introduction
The use of insect cells for the production of therapeutically active proteins has gained increasing importance over the last decade, highlighted by the market entry of the human papilloma virus vaccine Cervarix TM [1]. Nowadays, a set of different insect cell lines, suitable for the baculovirus-driven production of recombinant proteins, is available. The most popular ones are Spodoptera frugiperda Sf9 cells [2], Spodoptera frugiperda Sf21 cells [3] and Trichoplusia ni BTI-TN5B1-4 ''High Five'' (''Hi5'') cells [4]. More recently, a new Trichoplusia ni cell line, BTI-Tnao38, has been established [5]. A major limitation regarding the production of therapeutic proteins in insect cell lines is the lack of complex type N-glycans, often leading to a severely reduced efficacy. N-glycans found on insect cell expressed proteins are mainly of a high mannose type or non-fucosylated and core-fucosylated trimannose structures [6].
In the past, several approaches have served to improve N-linked glycosylation in lepidopteran insect cell lines by expressing additional glycosyltransferases. In a first trial, Sf9 cells were coinfected with standard baculovirus vectors expressing human b1,2-N-acetylglucosaminyltraferase I (GnTI) and a fowl plague virus haemagglutinin (HA). Analysis of HA glycosylation patterns gave a four fold increase in the amount of terminal N-acetylglucosamine (GlcNAc) indicating the proper functionality of human GnTI in insect cells [7]. In another approach, the use of a recombinant baculovirus expressing a bovine b1,4-galactosyltransferase driven by the immediate early promoter 1 (IE1) was investigated. It could be shown that the baculovirus envelope protein gp64 was modified by an additional galactose on at least one N-linked oligosaccharide side chain [8]. These strategies are based on transient expression of glycosyltransferases. Other studies have evaluated the use of stable integration for modifying insects N-linked glycosylation pathways. Hollister et al (1998) constructed a plasmid expressing a bovine b1,4-galactosyltransferase controlled by the baculoviral IE1 promoter and stably integrated it into the genome of Sf9 cells. A clone that has been selected for high enzymatic activity produced a galactosylated ''human tissue plasminogen activator'' protein after infection with the recombinant baculovirus [9]. The production of terminally sialylated biantennary N-glycans was the next step in glycoengineering insect cells. Several studies have shown that the stable integration of b1,4-galactosyltransferase and a2,6-sialylltransferase leads to partially sialylated gp64 after infection with the corresponding recombinant baculovirus [10,11]. To further extend the N-glycan processing, a set of five mammalian glycosyltransferases was integrated into the genome of Sf9 cells [12]. Due to the fact that the intracellular levels of sialic acid and CMP-sialic acid, necessary for proper sialylation, are very low in insect cells, a new transgenic Sf9 insect cell line was generated. This was achieved by the additional integration of genes encoding sialic acid synthetase and CMP-sialic acid synthetase leading to a total number of 7 constitutively expressed glycozymes [13]#. The drawback with using such a setup is a possible metabolic overload for the transgenic cell line, leading to reduced growth characteristics and long-term stability as well as reduced yields of recombinantly produced proteins. Furthermore, the altered glycosylation pattern might influence the functionality of cellular proteins and have a wider impact on the robustness of the system. To overcome these problems, a recent study has reported the generation of a new transgenic Sf9 cell line with an inducible mammalian-like protein N-glycosylation machinery [14]. What still remains is the fact that all these approaches are based on Sf9 cells. However, for the expression of secreted proteins, cell lines derived from Trichoplusia ni have been demonstrated to produce significantly higher amounts in many cases [15,16]. Therefore, we decided to revive the idea of using the baculovirus itself for glycoengineering purposes. The MultiBac platform, which is an advanced version of the Bac-to-Bac system, provides the possibility of flexible multigene expression using a single baculovirus vector [17][18][19][20]. In this study we evaluated the use of MultiBac for the generation of an efficient platform for the production of properly glycosylated proteins in lepitopteran insect cells other than the commonly used Sf9 cells.

Materials and Methods
All DNA manipulations were carried out essentially as summarised by Sambrook et al. [21]#. Restriction enzymes, T4 DNA ligase and Calf Intestinal Alkaline Phosphatase were purchased from New England Biolabs (Ipswich, USA). DNA polymerase was purchased from Novagen (Darmstadt, Germany). All enzymes were used according to manufacturer's recommendation. All primers and oligos were synthesised by Sigma-Aldrich (St. Louis, USA).

Construction of SweetBac
The Caenorhabditis elegans N-acetylglucosaminyltransferase II (GnTII) was PCR-amplified from a plasmid containing full length open reading frame (ORF) of the GnTII using primers CeGnTIIfor (59-GAC TTG ATC ACC CGG GAT GAT GGT CTA TCG ACG GAT G -39) and CeGnTII-rev (59-TGC ATC AGC TGC TAG CTT AAG AAG TTG TAG ATG TGA TTG T -39). The product was digested with NheI/XmaI and ligated into pUCDM vector (EMBL, Grenoble) cut with same enzymes, resulting in pUCDM-G. The bovine ß1,4-galactosyltransferase I (GalT) was PCR amplified from a plasmid containing full length ORF of the GalT using primers b4GalT-for (59-CAT CGG GCG CGG ATC CAT GAA GTT TCG GGA GCC GCT -39) and b4GalT-rev (59-GAC TGC AGG CTC TAG ACT AGC TCG GCG TCC CGA TG -39). The product was digested with BamHI/XbaI and ligated into pUCDM-G vector cut with the same enzymes resulting in pUCDM-GG. The resulting glycomodule was introduced in the loxP site of a MultiBac genome, using DH10MultiBac Cre cells according to Fitzgerald et al. [18]. The hence generated DH10MultiBac Glyco cells were then used as a basis for the generation of a recombinant baculovirus expressing an IgG 1 antibody.

Expression and purification of mAb 3D6
Tnao38 and Hi5 cells were transferred to shaker flasks with supplemented IP-L41 medium without FCS two days prior to infection. For infection, 5610 7 cells were spun down at 9006g for 10 min, resuspended in 50 mL fresh IP-L41 medium and infected with a multiplicity of infection (MOI) of 5. All expressions were carried out in triplicate. To generate an expression profile, samples were taken from 24 to 168 hours post infection (hpi) every 24 hours. For standardisation of further experiments, supernatants of Tnao38 and Hi5 cells were harvested 96 hpi by centrifugation at 2000 g for 10 min.
The cellular supernatant was filtered through a 0.22 mM filter cartridge (Millipore, Billerica, USA) and applied on a 1 mL Bio-Scale TM Mini UNOsphere SUPrA TM Cartridge (BioRad, Hercules, USA) using an Ä kta purifier (GEHealthcare). Elution was performed with 100 mM glycine/HCl buffer pH 3. To increase stability, the eluted samples were adjusted to pH 7 using 100 mM Tris-HCl (pH 9).

Enzyme-linked Immunosorbent Assay
Coating of 96 well Maxisorp plates (Nunc, Roskilde, Denmark) was performed by incubating 100 mL anti-Human IgG (c-chain specific) antibody (Sigma Aldrich, I3382) (diluted 1:1000 in coating buffer: PBS, 1% BSA) at 4uC over night. Plates were washed with TPBS (PBS, 0.1% Tween20) and 50 mL serially diluted samples (in dilution buffer: TPBS, 1% BSA) were applied to the wells. After one hour incubation on a rocking platform, the plates were washed again with TPBS. Detection was performed with 50 mL/well of an anti-human k light chain peroxidase conjugate (Sigma Aldrich, A7164) 1:1000 diluted in dilution buffer for 1 h while rocking. Staining of the ELISA plates was performed by adding 100 mL/well staining buffer (sodium citrate buffer, pH5) containing 1 mg/mL OPD (1,2-o-phenylenedihydrochloride) and 1 mL/mL H 2 O 2 (35%). The reaction was stopped with 100 mL H 2 SO 4 and plates were analysed on an InfiniteH M1000 reader (Tecan Group, Mä nnedorf, Switzerland).

N-glycan analysis
N-glycan analyses were essentially performed as described in Rendic et al. [24]. Briefly, after the heavy and light chains of purified IgG 1 were separated on an SDS-PAGE, bands corresponding to heavy chains were excised. After washing the bands, they were treated with dithiothreitol and iodoacetamide solutions to modify cysteine residues. The gel bands were washed again and then subjected to trypsin digestion at 37uC over night. The peptides extracted with AcN:H 2 O:TFA solution (acetonitrile/ water/trifluoroacetic acid; 666:333:1) were then dried, resuspended in 20 ml of 50 mM ammonium acetate (pH 5) buffer and subjected to PNGase A treatment at 37uC over night. The released N-glycans were separated from peptides by using columns packed with 10 mL LiChroprep RP 18 (25-40 mM) reversed phase resin (Merck) on top of 40 mL Dowex 50WX8-400 ion exchange resin (Sigma, 217514). After equilibrating the column with 2% acetic acid, the sample was applied, the column was washed with 2% acetic acid and the flow-through containing N-glycans was collected. For further purification of the released N-glycans, columns were packed with Supelclean TM ENVI-Carb TM PGC material (Sigma) on top of LiChroprep RP 18 (25-40 mM) reversed phase resin (Merck) and equilibrated with 2% acetic acid. The samples were loaded on the column and after washing with H 2 O, the N-glycans were eluted with 40% acetonitrile. The purified N-glycans were dried in a SpeedVac, finally resuspended in 5 mL deionised water and used for MALDI-TOF-MS analysis (on a Bruker Ultraflex MALDI-TOF/TOF in positive reflectron mode) using 6-aza-2-thiothymine (ATT) as matrix.

Flow cytometric analysis
Sf9 cells were infected with a recombinant baculovirus expressing the HIV gp41 3D6 epitope (SGKLICTTAVPWNAS) on the cellular surface with MOI 10 and harvested 72 hpi. After brief washing with phosphate buffered saline (PBS), 1610 6 cells were incubated with 100 mL of mammalianised/''wild-type'' 3D6 antibody (1 mg/mL in PBS containing 10% FCS) for 1 h at room temperature while gently shaking. Cells were again washed with PBS and incubated with an anti-human IgG (c-chain specific) Rphycoerythrin conjugate (Sigma-Aldrich, P8047) in PBS containing 10% FCS for 1 h at room temperature. After a final wash with PBS, cells were resuspended in 400 mL PBS and analysed on a FACS-CantoII (Becton Dickinson, San Jose, CA).
Human U937 leukemic monocyte lymphoma cells were washed with PBS and incubated with 100 mL of mammalianised/''wildtype'' 3D6 antibody (1 mg/mL in PBS) for 1 h at room temperature while gently shaking. After a brief washing, cells were incubated with an anti-human k light chain FITC conjugate (Sigma Aldrich, F3761). Cells were finally washed with PBS, resuspended in 400 mL PBS and analysed on a FACS-CantoII.

Construction of SweetBac
Our aim was to develop a virus backbone, facilitating flexible modifications of the N-linked glycosylation capacity of Sf9, Sf21, Hi5 and Tnao38 insect cells. Previous results have shown that mammalian b1,4-galactosyltransferase I (GalT) can efficiently modify the glycosylation of insect cells [9]#. Our own data further indicate that the invertebrate GLY-20 N-acetylglucosaminyltransferase II (GnTII) of C. elegans, whose temperature optimum is near to that where insect cells are grown, was well expressed in Sf9 cells (Wang H., Rendic D., unpublished data). Therefore, we constructed a glyco-module coding for a C. elegans GnTII and a bovine GalT controlled by p10 and polyhedrin (ph) promoter respectively. This module was integrated into the loxP site of a MultiBac genome. The resulting SweetBac was then used as a basis for the production of recombinant glycoproteins with mammalianised complex type N-glycans. In order to evaluate the functionality of SweetBac, we additionally introduced heavy (HC) and light chain (LC) open reading frames of the human HIV anti-gp41 antibody 3D6 [22] in the Tn7 site of a SweetBac genome, resulting in SweetBac-3D6. As a negative control, open reading frames coding the heavy and the light chains of the same antibody were integrated in a ''wild-type'' MultiBac genome as well (Figure 1). Previous studies have demonstrated higher yields, especially for secreted proteins, when using Hi5 cells [15]. In order to evaluate antibody expression yields for both cell lines, we infected Tnao38 and Hi5 cells with a recombinant baculovirus expressing 3D6 antibody at an MOI of 5 and took samples every 24 hours up to 4 or 7 days; Hi5 cells started to lyse after 4 days and thus, were not cultivated any longer. The amount of secreted antibody was measured by ELISA. Figure 2A shows that the secretion profile among the two cell lines was similar within the first 96 hours post infection (hpi). Tnao38 and Hi5 cells yielded around 1 mg of secreted antibody/ml cell culture supernatant 24 hpi, steadily increasing to 9.5 and 12 mg respectively at 96 hpi. Tnao38 cells were more stable, allowing the expression to be expanded up to 168 hpi. A continuous increase in antibody secretion could be detected, yielding up to 30 mg/ml on day 7. These data indicate that Tnao38 cells are even more robust than Hi5 cells and suggest them to be, at least for this antibody, highly suitable for large scale protein production.
To evaluate whether co-expression of the glycosyltransferases has an impact on target gene expressions, we infected Tnao38 and Hi5 cells with a SweetBac virus expressing the 3D6 antibody (SweetBac-3D6) and with standard baculovirus expressing 3D6. Samples were analysed at 72 hpi and 96 hpi for both cell lines. Figure 2B shows that at 72 hpi, the amounts of secreted 3D6 antibody from Tnao38 and Hi5 cells decreased 18% as compared to 3D6 expression without glycosyltransferase genes in the baculovirus genome. At 96 hpi, the yields were 25% lower for Tnao38 cells and 30% lower for Hi5 cells when SweetBac-3D6 was used. Significance of the results was confirmed by Student's t-test (p-value,0.05).

Testing the glycosylation of 3D6
For testing the functionality of SweetBac, purified 3D6 antibodies were subjected to SDS-PAGE and electroblotted to a nitrocellulose membrane. The presence of galactose residues was detected by incubating the blot with Ricinus communis agglutinin I (RCA I), specific for terminal galactose [25]#. Figure 3 shows that although similar protein amounts of antibody heavy chains were loaded on the gel, RCA I bound to only mammalianised variants (HiGlyco and Tn38Glyco). These results confirmed the desired change in glycosylation, indicating proper functionality of the integrated glycosyltransferases. 3D6 derived from Chinese hamster ovary cells (CHO) as well as one derived from Mimic Sf9 cells, a cell line containing a human N-acetylglucosaminyltransferase I, human N-acetylglucosaminyltransferase II, rat b1,4-galactosyltransferase I, mouse a2,3-sialyltransferase and mouse a2,6sialyltransferase [12] served as controls.

N-glycan analysis
In order to confirm the results obtained with lectin blotting, we analyzed the N-glycosylation patterns of the different 3D6 antibodies by mass spectrometry. Figure 4 shows the MALDI-TOF-MS spectra of PNGase A released IgG 1 N-glycans from mammalianised and ''wild-type'' Tnao38 and Hi5 cells. The 3D6 expressed in Tnao38 cells carries mainly single and double fucosylated tri-mannose structures, partly with one terminal Nacetylglucosamine. 3D6 antibodies from mammalianised Tnao38 cells on the other hand showed mainly complex, singly fucosylated biantennary N-glycan structures carrying two terminal N-acetylglucosamine residues (,30%), one terminal galactose (,30%) or two terminal galactose residues (,20%). N-glycans from Hi5 cell expressed 3D6 closely resemble those found from Tnao38 cells, but are more highly galactosylated. The portion of complex biantennary structures was as high as ,80% from mammalianised variants with ,50% carrying two terminal galactose residues.

Functionality of 3D6
In order to evaluate the target binding ability of the different 3D6 antibodies, we infected Sf9 cells with a recombinant baculovirus expressing the HIV gp41 epitope (SGKLICT-TAVPWNAS) on the cellular surface. The epitope-presenting cells were incubated with 3D6 antibody and analysed by FACS. Figure 5A shows that the binding characteristics were equal for all antibody preparations, indicating the same specificity and affinity.
Besides target binding, the ability of an antibody to elicit glycosylation dependent effector functions was evaluated. Human U937 leukemic monocyte lymphoma cells that are known to express the human Fc gamma receptor I (CD64) on their cellular surface were incubated with 3D6 antibody samples and receptor binding was again analysed by FACS. Figure 5B shows that binding characteristics of antibodies produced in Hi5 and Tnao38 cells were comparable. However, the peaks resulting from samples derived from SweetBac-3D6 infected cell lines are shifted in both cases. This increase in binding capacity indicates the importance of proper, mammalian-like N-glycosylation for recombinantly expressed antibodies [26][27][28].

Discussion
The presence of specific N-linked glycan structures is of great importance for the full functionality of therapeutically active proteins. Mammalian cells, especially Chinese hamster ovary (CHO), human embryonic kidney (HEK) and mouse myeloma (NS0) cells, are currently the favoured cell lines for the production of functional proteins like antibodies, because of high yields and their ability to generate complex type N-glycans [29,30]. A major drawback in the field of mammalian cell technology is the either time consuming top-clone selection for generating transgenic cell lines, the labour-intensive transfection procedures for transient expressions or, occasionally, the occurrence of allergenic a1,3galactose [31]. The baculovirus-insect cell system may have an advantage over stably transfected CHO cells, when considering a need for fast and easy production systems which may get increasingly important in the future due to novel, ''personalised'' medical treatments or when time consuming production schemes are not acceptable.
Based on data proposing that Hi5 cells are more suitable for expressing high amounts of secreted proteins as compared to the commonly used Sf9 cells [15,16], we chose to evaluate the expression capacity of Hi5 cells and the newly established Tnao38 cell line [5]#. Using Tnao38 cells, the expression could be expanded up to 7 days post infection because of very high cellular stability. In the end, the amount of secreted 3D6 antibody was as high as 30 mg/mL cell culture supernatant, which is approaching production levels of transient expressions in mammalian cells (Figure 2). The higher robustness of this cell line might be an attractive feature for up-scaling the production system. The less virulent infectious cycle is also reflected by 1-2 magnitudes lower baculovirus titres in the supernatant as compared to Sf9 cells (data not shown). The lower baculovirus titres in the supernatant are expected to reduce issues in downstream processing facilitating quicker and more effective steps towards acquiring pure (and virus-free) recombinantly produced protein.
In this study we further evaluated the possibility to mammalianise the insect glycosylation machinery by a viral-based approach in order to provide a flexible protein production platform suitable for various insect cell lines. The differences in N-glycan processing between insect and mammalian cells are due mainly to insignificant expression of glycosyltransferases respon-sible for generating complex-type structures. Glycoengineering in insect cells is based mainly on the pioneering work of Jarvis and co-workers and is generally performed by stably integrating enzymes into the cellular genome of Sf9 cells [12,13]. This, however, is a very time consuming process and a flexible choice of the most suitable cell line is lost. In a previous study we have furthermore shown that Sf9 cells are not always the ideal choice when it comes to the production of secreted proteins in high yields [16]#.
Our approach to provide a more flexible glycosylation to a wide range of insect cell lines is based on the MultiBac technology [17][18][19][20] which is feasible for the expression of multiple genes from one baculovirus backbone. We integrated the C. elegans N-acetylgluco- saminyltransferase II (GnTII) and the bovine b1,4-galactosyltransferase I (GalT) into the backbone of a standard MultiBac virus, resulting in a recombinant baculovirus designated as SweetBac. We further inserted heavy and light chain genes of the human HIV anti-gp41 IgG 1 antibody 3D6 [22] as a model, resulting in the baculovirus vector SweetBac-3D6. A great advantage of this setup is that the enzymes necessary for glycosylation as well as the genes of interest are present on one single baculoviral vector. Co-expression of proteins in one cell by co-infection of two or more recombinant baculovirus clones is generally possible. However, the use of multiple baculovirus clones for co-infection decreases the chances that all required genes are delivered to all cells in the population.
Studies have already reported the expression of glyco-modifying enzymes and target protein from the same viral vector, however, integration was based on a quadruple transfer-vector with a size of 16.6 kb [32]#. This approach limits the possibility of integrating further enzymes and/or target proteins. The reason for initially focussing on two glycosyltransferases was the fact that proper sialylation requires a whole set of enzymes and thus, is beyond the scope of demonstrating the proof of concept [13]. However, the existing SweetBac virus may further be modified by integrating respective expression cassettes into its backbone.
When SweetBac-3D6 was used to infect Hi5 and Tnao38 cells the yield of antibody was significantly reduced as compared to expression without a glycosylation module. This might be due to the very strong promoters used for the expression of glycosyltransferases. Using the weak immediate early promoter IE1, providing proper glycosylation earlier during the infection cycle, might have a positive impact on the product as well.
In order to evaluate the actual change in glycosylation of 3D6, we performed a lectin blot detecting terminal galactose. Figure 3 shows that specific binding was only detected with mammalianised antibodies indicating a proper functionality of the introduced glycosyltransferases. As a conformation, we analysed PNGase A released N-glycans by MALDI-TOF-MS. Indeed, SweetBac-3D6 Figure 5. Binding of insect cell expressed 3D6 antibodies to 3D6 epitope and human FccRI. Sf9 cells presenting the 3D6 epitope were incubated with purified antibodies and target binding was analysed by FACS. Antibodies expressed in Hi5 and Tnao38 cells using standard baculoviral vectors (Hi5, Tnao38) as well as SweetBac (Hi5 Glyco, Tnao38 Glyco) show a highly specific binding. The broadening of the peaks can be explained by different amounts of 3D6 epitope displayed on Sf9 cells (A). Binding of antibodies to human Fc gamma receptor I was measured by incubating 3D6 antibodies with U937 cells. FACS analysis showed that all antibodies bound FccRI present on the cellular surface, but the binding of SweetBac expressed variants (Hi5 Glyco, Tnao38 Glyco) was significantly increased. doi:10.1371/journal.pone.0034226.g005 derived antibody carries complex biantennary N-glycans, mainly with one or two terminal galactose residues. Nevertheless, the amount of galactosylated glycans on IgGs expressed in mammalianised Hi5 cells is significantly higher (,50% compared to ,20% in Tnao38 cells) (Figure 4). In terms of the degree of galactosylation, SweetBac mammalianised Tnao38 cells produced IgG akin to that observed for CHO cells [16]. However, Tnao38 cell derived IgGs also carried more paucimannosidic structures, lacking terminal GlcNAc or galactose; these types of glycans are only present at low levels on IgGs expressed by Hi5 cells and are absent from CHO cells.
It is known that the N-linked glycosylation of an IgG 1 antibody strongly influences the ability to exert effector functions like antibody-dependent cellular cytotoxicity (ADCC). We therefore investigated the binding of 3D6 to human Fcc receptors present on human U937 leukemic monocyte lymphoma cells, as well as the binding to its target. The presence of a more complex Nglycan structure did not influence target binding ( Figure 5A), but binding to Fc receptors was significantly enhanced ( Figure 5B), indicating that one can improve the efficacy of insect cell expressed antibodies by co-expressing two key glycosyltransferases using the SweetBac system. As insect cell expression system has been gaining more importance during the past years [1], and new cell lines are being established [5], we aim to provide a flexible tool for production of custom-designed individual products at sufficient yields and within a reasonable time frame.