Cytokine mRNA and protein expression by cell cultures of epithelial ovarian cancer—Methodological considerations on the choice of analytical method for cytokine analyses

To get a comprehensive picture of cytokine expression in health and disease is difficult, cytokines are transiently and locally expressed, and protein analyses are burdened by biological modifications, technical issues, and sensitivity to handling of samples. Thus, alternative methods, based on molecular techniques for cytokine mRNA analyses, are often used. We compared cytokine mRNA and protein expression to evaluate whether cytokine mRNA profiles can be used instead of protein analyses.


| INTRODUC TI ON
Recently, the involvement and importance of the immune system in cancer and its treatment have righteously and finally been acknowledged as an essential factor, playing a decisive role on the curability, prognosis and outcome of this disease. The research field of oncoimmunology, studying the interactions of tumor cells and the immune system, has achieved a great progress in understanding the patho- Cytokines are produced and secreted by a variety of cells to coordinate the immune response and provide growth-, differentiation-, inflammatory-, and immunosuppressive signals. The released cytokine/group of cytokines following a certain stimuli determine(s) the subsequent immune response. 4 The cytokine signal is potentiated and diversified by cascades-a cytokine stimulates the production of the same or other cytokines and/or cytokine receptors, further potentiating and forwarding the prevailing immune response. 3 It has been shown that different cytokine mRNA profiles, designated Th1, Th2, Th3/Tr1, and Th17, are associated with the ability to mediate and regulate immunity and inflammation, promote or halt growth, movement or immune responses. Thus, a cytokine profile dominated by IFN-γ, IL-12, and IL-15 (Th1) promotes cytotoxicity, a cytokine profile dominated by IL-4, IL-5, and IL-13 (Th2) promotes humoral immunity, IL-1β, IL-6, IL-8, IL-17, TNF-α, and TNF-β/LTA promote inflammation and TGF-β1 and IL-10 (Th3/Tr1) promote immunosuppression, and innate and adaptive T regulatory cell development. 5 Since cytokines are produced and act locally, with a short halflife, 5 they are often only transiently present in small amounts and/ or absent, as proteins, at a given moment in a biological system.
Because of this, a need for alternative methods for revealing cytokine presence has emerged. Real-time quantitative RT-PCR is one of the methods often used as a proxy to cytokine protein analyses. [6][7][8][9][10][11][12] It has also been an important technique in identifying different diagnostic tests and predicting outcomes for different malignant and non-malignant diseases. 13 Recently, we used this method to gain knowledge on the cytokine mRNA expression profile in paired ovarian cancer tissue samples and peripheral blood mononuclear cell (PBMC) samples of women suffering from HGSC. 6 Our results indicated elevated cytokine mRNA levels in the ovarian cancer microenvironment compared with non-malignant conditions and normal ovarian tissue, where inflammation, immune suppression, and promotion of T reg cells prevailed.
It is well known and proven that not all mRNA signals are translated to proteins. 14,15 Assessing mRNA levels exclusively will only give information on transcribing DNA to mRNA and not on the further protein production and secretion. In addition, cytokines have a diverse range of action, some showing de novo synthesis and others being stored intracellularly and released immediately upon stimulation. Considering the instability of cytokine proteins in serum, it could be of interest to study if it is sufficient to analyze cytokine mRNA profiles to gain knowledge on a specific immune response, and in what instances cytokine mRNA could be used as a proxy for protein analysis.
Here, we compared the mRNA expression for a set of cytokines with the corresponding extracellular protein expression, using quantitative RT-PCR and multiplex protein analysis (Luminex ® ) in kinetic experiments. Traditionally, enzyme-linked immunosorbent assay (ELISA) has been used for cytokine protein analyses. The advantage of multiplex bead array assays, such as Luminex ® , is its efficiency as a high throughput multiplex method analyzing several cytokines at the same time. The sensitivity for each method is comparable. 16,17 In kinetic experiments, we assessed cytokine mRNA and protein expression in three cell lines after activation by thermal stress-the epithelial ovarian cancer cell lines OVCAR-3 and SKOV-3, and Jurkat, a T-cell-derived cell line that was used as a control to the ovarian cancer cell lines. Time points were chosen to capture cytokine mRNA and protein production peaks. 18 Working with cell lines presents a unique opportunity to control the experimental conditions. The correlation between cytokine mRNA and protein expression in kinetic experiments has to our knowledge not been previously investigated in ovarian cancer cells.

| Activation of cytokine gene expression by thermal stress
Confluent OVCAR-3 and SKOV-3 cultures in 25 cm 2 culture flasks (VMR international), and 3 × 10 6 /mL Jurkat cells were incubated at 42°C for 1 hour to stimulate cytokine expression. Successful heat shock effect was confirmed by assessing increased mRNA levels of HSP4A, MICA, and MICB ( Figure 1) by RT-qPCR. Cell culture supernatant and cells were collected for further analyses at seven consecutive time points; at starting point 0, before heat shocking, and at 1, 3, 4.5, 6, 9, and 24 hour post-heat shocking of the cells.

| Patient samples
Serum samples for cytokine protein assessment of 14 women suffering from HGSC were retrieved from the Ovarian Cancer Biobank at Norrland's university hospital. The samples were collected after F I G U R E 1 Effect of heat shock on thermal stress-inducible genes HSP4A, MICA, and MICB. mRNA upregulation at seven consecutive time points for the ovarian cancer cell lines OVCAR-3 and SKOV-3, and the T-cell line Jurkat, used as a control ethical permission and informed consent. Cytokine mRNA expression in ovarian tumor tissue from the same patients, previously studied and published, 6 was used as comparison to the protein cytokine assessment.

TA B L E 1
Average Ct values ± standard deviation for the mRNA expression of the cytokines and thermal stress-inducible genes, and the endogenous control gene 18S rRNA, in OVCAR-3, SKOV-3 and Jurkat, before and at different time points after thermal stress (n = 3)

| Statistical analyses
Spearman's rank order correlation was used to assess the correlation between cytokine mRNA and protein expression at the different time points. The IBM SPSS software version 25 was used for these analyses. Since the time points are not more than seven, the P-values are not given. in Jurkat, where the second peak at 9 hour was higher than the initial peak at 1 hour and MICA in SKOV-3, where the highest value was seen at starting point. From these experiments, we concluded that the experimental set-up of thermal stress conditions gave expected and satisfactory results and we could proceed with the cytokine analyses.

| RE
The mean individual Ct values and SD of each cell line, internal control, type of cytokine, and time point can be seen in Table 1.   ( Figure 2B and C). As can be seen, the initial high peak was followed by a deep fall at 3-9 hours. After that, a continuous upregulation of the mRNA signal was seen, but at 24 hour not reaching the level of the initial peak. Both IL-6 and IL-8 were detected at protein level in SKOV-3 culture supernatants ( Figure 2C). Both cytokines exhibited the highest protein concentration at starting point, 9.5 pg/mL for IL-6 and 1230.8 pg/mL for IL-8, and then decreased and reached a steady state at about 5 pg/mL and 400 pg/mL, respectively.  Table 1.

| Ovarian cancer cells express cytokine
The mean and SD of the protein expression analyzed by Luminex ® are shown in Table 2.

| Translation of cytokine mRNA to proteins can be missed due to dilution under the protein detection limit of the assay
The detection of cytokine proteins in serum can be biased due to the fact that they are produced and act locally in small amounts, it is only during a high production they will "spill over" to the peripheral blood.
Reaching the blood and or other body fluids, they would be diluted and, in some instances, would not be detected if they are below detection level of the analytical method used. The result can thus be wrongly interpreted as a failure to translate cytokine mRNA to protein. The dilution effect was also illustrated in our kinetic experiments. Thus, TNF-α protein expression by OVCAR-3 was revealed after 6× concentration of the used culture supernatant ( Figure 2B).
The same procedure was done to reveal IL-8 and TNF-β/LTA protein expression in supernatants from Jurkat cultures ( Figure 2D). To test whether changing the concentration of the used supernatant could alter the correlation between mRNA signal and protein expression, we analyzed IL-6 in OVCAR-3 and SKOV-3 in non-concentrated and 6× concentrated used supernatants (Figure 3) and found a similar curve pattern for protein expression.

| Comparison of IL-8 and IL-6 mRNA and protein analyses in paired tumor and serum samples in HGSC patients
We analyzed, in paired samples, the protein expression of IL-1β, IL-6, IL-8, TNF-α, and TNF-β/LTA in the serum of 14 women, suffering from HGSC, and compared it to previously analyzed mRNA expression in biopsies from their tumors. 6 The results are summarized in   (Table 3).
Since the investigated time points are few in the kinetic experiments, and the serum samples are taken at only one occasion, no conclusions can be drawn on statistical significance.     parison has to be made on a much larger patient material before conclusions can be drawn on replacing cytokine protein analysis by mRNA analysis when used specifically for gaining knowledge on protein level and function. But, for comparisons between sample groups (for example benign vs malignant) where the emphasis is on the difference between groups and not in an exact individual protein level, we conclude that analyzing cytokine mRNA profiles by RT-qPCR is more reliable in detecting very low abundance of biological molecules, with a wider detection span, and less affected by post-sampling handling. In addition, if no concrete protein values are needed, immunohistochemistry could be considered as a useful complement to mRNA expression analysis to provide a semi-quantitative picture of protein levels.
In conclusion, so far, the comparisons between cytokine mRNA and protein expression levels performed on human material show very diverging results, often specific to a certain material (PBMC/ tissue/serum/plasma) and method (PCR/microarray/ELISA/flow cytometry/multiplex immunoassay/polyacrylamide gel electrophoresis) making it difficult to apply outside the investigated system. [23][24][25][26] Considering the very high sensitivity and reproducibility of the real-time quantitative RT-PCR method, we would like to suggest that determination of cytokine mRNA profiles could be used as a proxy for protein-mediated functions for some specific purposes, such as comparisons between different patient groups and in defining mechanistic pathways involved in the pathogenesis of cancer and other pathological conditions.

ACK N OWLED G M ENTS
We are grateful to all donors, the staff at the Department of