Raw Material Characterization: Effect of Trace Metal Variability on Protein Glycosylation

By: Scott Wilson, Barry Drew, Joanna Grayham, and Chandana Sharma SAFC, Cell Sciences and Development, 13804 W 107th St, Lenexa, KS 66062 USA


Trace metal variability has been observed in cell culture products. SAFC’s raw material characterization program has identified that variability is typically caused by raw materials used in the manufacturing of these products. Even differences seen at ppb concentration can still potentially impact a biological system. Using an intact mass glycosylation assay, dose response, and Design of Experiment screening with an industry relevant CHO cell line, SAFC has evaluated the impact of 15 trace metals on our biological systems critical protein quality attributes. Statistical and multivariate analysis has been used to confirm the significance of any observed impact. The analysis has identified Fe and Mn at high concentrations can reduce the percent G0F and increase the G1F and G2F glycoforms. The effect of other trace metals on glycosylation was determined to be insignificant using ANOVA analysis (p values >0.05).


Understanding protein glycosylation is very important in the development of therapeutic proteins. Any changes that may occur to these proteins’ critical quality attributes can result in difference in the products efficacy and safety1. It has been well studied by the industry that specific metals can alter the glycosylation profile, thus it is important for media manufactures to fully understand trace metal variability and its effect2. SAFC’s data suggests that the metal variability comes from the raw materials used in the finished product manufacturing. These raw materials typically contain trace metals ranging from ppb to ppm concentrations. SAFC’s studies are designed to determine if the very low concentrations contributed by these raw materials would effect protein glycosylation profiles.

Biological Method

Screen 1:

Fifteen metals were evaluated in a high throughput 96 deep well plate dose response cell culture assay (Table 1). Eight different concentrations were screened for each of these metals using a CHO Zn™ K1 GS knock out cell line. The 96 deep well plates were cultured at 37°C for seven days with samples pulled days 3, 5, and 7 to analyze cell growth. A PromegaCellTiterGlo® kit was used to correlate relative luminescent units to viable cell density. On day 7, all cultures were harvested and protein concentration was quantitated using a Forte Bio Octet QK. Harvested samples were also analyzed for glycosylation profile differences.

Screen 2:

Data generated from screen 1 was used to eliminate metals that had no effect on the glycosylation profile. Therefore only ten metals were included as part of screen 2. Experimental design for screen 2 was developed using Stat-Ease Design Expert® software(DOE). DOE allowed for model designs that could identify metal effect significance as well as metal-to-metal interactions. The DOE would be cultured in a TPP Bioreactor tube system, which allowed for greater culture volume. The increased culture volume allowed for glycosylation to be tested on days 5, 7, and 10. Once the experiments were complete, the data was analyzed using DOE software as well as Umetrics SIMCA multivariate analysis.

Analytical Method

The proteins were first purified using protein A affinity chromatography. These purified proteins were then analyzed for intact mass glycosylation using a Waters Aquity UPLC with a AquityUPLC PrST SEC column and Xevo qToF mass spectrometer. Data was then processed using Biopharmalynx software to quantitate percent G0, G0F, G1F, G2F, non-glycosylated, and Man5 glycoforms.

Results and Discussion

Screen 1:

Table 1 summarizes each metals effect on protein glycosylation. Six of the 15 metals tested had no effect on protein glycosylation. Seven metals had slight differences that were not considered significant. More evaluation was necessary for these metals to prove any possible differences. Iron and manganese had significant effect on G0F, G1F, and G2F glycoforms (Figure 1 and 2).

Glycosylation Effect Trace Metal
No Effect I, J, K, L, M, N
? B, D, C, F, G, H, E
Glycosylation Effect Fe, Mn, A

Table 1. Summary of metal effect on glycosylation (Trace Metals are coded A-N). Significant effect was seen with Fe
and Mn. The question mark indicates any difference seen was not significant, but more data was needed to confirm results.



Figure 1. High concentrations of iron leads to change in glycosylation profile. As iron is increased percent G0F decreases and G1F and G2F increase. Data lower than 2ppm Fe was removed due to low protein concentration.


Figure 2. Differences were also seen at high concentrations of manganese. As manganese concentration increases percent G0F decreases while G1F and G2F both increase.

Screen 2:

All metals that showed no glycosylation impact during screen 1 were eliminated for the second screening. A DOE model showed significance for three trace metals with p-values <0.05. Trace metal A, Fe, and Mn all impacted the glycosylation profile (Table 2). The data suggests that no interactions between metals occurred that caused significant changes to the glycosylation profile. Multivariate analysis (Figure 3) confirmed the differences seen during screen 1 as well as identified that trace metal A had significant impact on glycosylation. If trace metal A is increased, percent glycoform G0F increases.

Table 2. ANOVA analysis from the DOE model suggests that trace metal A, Mn, and Fe significantly effect protein glycosylation. All p-values <0.05 are marked with red value indicating significance.



Figure 3. A PLS Loading Scatter Plot show correlation between the x variables (metals) and response y variables (% glycoforms). The plot suggests as Fe and Mn increased the G1F and G2F glycoforms increase.


  • Fe, Mn, and Trace metal A impact protein glycosylation profiles
  • Fe and Mn increased percent G1F and G2F glycoforms and decreased G0F
  • Trace Metal A increased G0F while decreasing G1F and G2F
  • In SAFC’s cell culture system only high concentrations had impact on glycosylation. Impact on seen at > 6 ppm for Fe and >20 ppb Mn.
  • Differences at low concentrations did not impact glycosylation. Differences at < 6ppm for Fe and <20ppb Mn were with in the assays variability.
  • In development of a biologic therapeutic it is necessary to understand the systems tolerance for metal variability.



References and Acknowledgments

  1. Scientific Considerations in Demonstrating Biosimilarity to a Reference Product Guidance for Industry; U.S. Department of Health and Human Services, (US) Food and Drug Administration, (US) Center for Drug Evaluation and Research, (US) Center for Biologics Evaluation and Research (US): Silver Spring, MD, 2015 .
  2. Pacis, E.; Yu, M.; Autsen, J.; Bayer, R.; Li, F. Effects of Cell culture conditions on antibody N-lined glycosylation – what affects high mannose 5 glycoform. Biotechnol Bioeng. 2011 Oct; 2348-58


Related Links