January 1, 2014 (Vol. 34, No. 1)

Abhishek J. Gupta Product Developer FrieslandCampina Domo
Kathleen Harrison Senior Scientist FrieslandCampina Domo
Dominick Maes R&D Manager FrieslandCampina Domo

Range of Compounds Impacts Cell Growth and Antibody Production in Mammalian Cell Lines

Soy protein hydrolysates are commonly used in mammalian cell culture. They contain a wide variety of compounds such as peptides, free amino acids, carbohydrates, lipids, organic acids, and nucleotides. This complex composition of hydrolysates results in enhanced mammalian cell growth and production.

FrieslandCampina Domo is studying how different compounds present in hydrolysates affect cell growth and antibody production in mammalian cell lines.


Untargeted metabolomics approach and cell culture analysis
Proyield Soy SE50MAF-UF, an ultrafiltered soy protein hydrolysate produced by FrieslandCampina Domo, was chosen for this study. Thirty samples, specifically selected to have the widest variation in cell culture functionality, were analyzed using an untargeted metabolomics approach. Hydrolysates were analyzed using liquid/gas chromatography coupled to mass spectrometry. All compounds detected in Proyield Soy SE50MAF-UF hydrolysates were semi-quantified using mass spectrometry.

In addition to metabolomics analysis, all the hydrolysates were examined in cell culture functionality tests to determine IgG production, which was measured using the ELISA method.

Data pretreatment
The semi-quantitative composition data of Proyield Soy SE50MAF-UF was natural log transformed to correct for the fact that it was not normally distributed. The corrected composition and the IgG production data were integrated and analyzed further using three in-house developed Matlab tools (Matlab® R2009b version The tools were two-mode cluster analysis, 2D correlation maps, and bootstrap stepwise regression.

Two-mode cluster analysis
The descriptive statistical technique known as two-mode cluster analysis is based on the principle of two-mode k-means clustering. The optimal hydrolysate-compound cluster combination needed to run two-mode cluster analysis was derived from a knee plot. Using this combination, that is, six hydrolysate clusters and eight compound clusters, the compounds, hydrolysates, and the IgG production data were clustered simultaneously. It showed which compounds behaved similarly for which hydrolysates and determined the compounds that were related to IgG production.

2D correlation maps
By means of 2D correlation maps, the correlation coefficients between IgG production and compounds were calculated. The significant correlations (p-value < 0.05) and correlation coefficients greater than or equal to the threshold correlation coefficient (−1 to +1) were selected to make the correlation plots.

The plotting was done using classical multidimensional scaling where the higher dimensional distances between the compounds are retained and plotted in a 2D space. The thickness of the lines in the plots indicated the strength of the correlations. Whereas a red colored line indicated a negative correlation, blue colored lines showed positive correlations.

Validation assay
To validate the statistical findings, the key compounds that were commercially available were purchased and tested in cell culture experiments. The compounds tested were ferulic acid, syringic acid, galactarate, adenine, and trigonelline. The quantification of these compounds was limited because of the interference by the complex matrix of hydrolysates.

Several patents on cell culture medium formulations for mammalian cells already describe the use of inositol, adenine, and nicotinamide (trigonelline-related compounds) in the ranges 0.0005–0.002%, 0.000001–0.0075%, and 0.0001–0.06%, respectively.

Additionally, use of lactate (0.001%) in Dulbecco’s Modified Eagle’s Medium for cell culture is known in literature. Thus, it was hypothesized that the chemically defined medium used in this study already contained some key compounds identified in this study. Additionally, based on the positive correlations found between the key compounds and IgG production, the compounds were added at 0.01% (w/v) concentration to chemically defined medium, and cell culture assays were performed.

Chemical Composition

Of the 410 compounds detected in Proyield Soy SE50MAF-UF hydrolysates, 157 were successfully annotated, and the remaining 253 were tentatively assigned as peptides. There were 97 compounds that were found in some but not all of the 30 hydrolysates, and therefore they were removed from the dataset. The remaining 313 compounds were included in the final data analysis.

Cell Culture

All Proyield Soy SE50MAF-UF supplemented cell cultures had significantly higher IgG production (117–283%) than cultures grown in chemically defined medium (100%). In some Proyield Soy SE50MAF-UF supplemented cultures, the IgG production was even higher than that obtained in fetal calf serum supplemented cultures.

This shows that when hydrolysate quality is controlled, the functionality of Proyield Soy SE50MAF-UF can be enhanced even further.

Two-Mode Cluster Analysis

The hydrolysates were grouped in six clusters (y-axis) and the compounds were grouped in eight clusters (x-axis) in the two-mode cluster analysis. The red, green, and black colors indicated a high, low, or an average relative concentration of the compounds in hydrolysates, respectively.

The analysis gave a descriptive qualitative overview of the compositional differences between the hydrolysates grouped in different clusters (Figure 1).

In two-mode cluster analysis, the IgG production was grouped in compound cluster 5. This suggested that the compounds present in this cluster were positively correlated to IgG production. For example, the hydrolysates in sample cluster D were high in concentration of cluster 5 compounds and IgG production, while sample cluster B was low in concentration of cluster 5 compounds and IgG production. This was further analyzed using 2D correlation maps.

Figure 1. In two-mode cluster analysis, IgG production was grouped in compound cluster 5, suggesting that the compounds in this cluster were positively correlated to IgG production. Note that the hydrosylates in sample cluster D have a higher concentration of cluster 5 compounds than the hydrosylates in sample cluster B.

2D Correlation Maps

Both compound-compound and compound-IgG correlations were observed at the same time (Figure 2).

This gave an insight into the complex interactions between compounds that occur within the hydrolysate and showed how these interactions influenced IgG production. Moreover, this suggested that it is not one compound but an entire network of compounds that contributed to the IgG production. To confirm this, cell culture experiments were performed in which pure compounds were supplemented in chemically defined medium.

Figure 2. With the use of 2D correlation maps, both compound-compound and compound-IgG correlations could be observed at the same time. The analysis suggested that not one compound but an entire network of compounds contributes to IgG production.

Validation Assay

The IgG production in Proyield Soy SE50MAF-UF supplemented culture was the highest. All the test compounds increased the IgG production as compared to chemically defined medium (control) (Figure 3).

Although there was an increase in IgG production due to key compounds, it was not to the same extent as that of Proyield Soy SE50MAF-UF. This confirms again that individual compounds present in Proyield Soy SE50MAF-UF act as biomarkers and the functionality of a hydrolysate is a result of the complex network of compounds.

Figure 3. In a validation experiment, several key compounds caused an increase in IgG production. None, however, caused as large an increase as that observed with Proyield Soy SE50MAF-UF, suggesting that the hydrosylate’s functionality results from a complex network of compounds.


Soy protein hydrolysates contained both proteinaceous and nonproteinaceous compounds, which were interconnected in a complex network. This unique feature of hydrolysates has now been analyzed in detail. It proved to be possible to identify certain marker compounds in the hydrolysate composition which correlate with variability in the cell culture applications.

Addition of these key compounds to media in some cases improves the titer, but the effect is always much lower than the overall effect of the complete hydrolysate. This suggests that the effect of a hydrolysate on specific as well as total IgG production cannot be reproduced by adding individual metabolites.

This research has given FrieslandCampina Domo detailed insights in the composition and variability of Proyield Soy SE50MAF-UF. The knowledge will be applied to further improve the performance and consistency of Proyield Soy SE50MAF-UF and future products for mammalian cell culture applications.

Abhishek Gupta is a product developer, Kathleen Harrison (kathleen.harrison@frieslandcampina.com) serves as senior scientist-Biopharma Ingredients, and Dominick Maes is R&D manager-BioPharma Ingredients for FrieslandCampina Domo.

This site uses Akismet to reduce spam. Learn how your comment data is processed.