Unfavorable Effects of Fixatives on the Fluorescence Intensity and Biological Functions of Fluorescent Proteins in HEK293T Cells and Transgenic Mice

Fluorescent proteins (FPs) are commonly used probes for coding genes that enable specic protein or whole-cell labeling. Their uorescence intensity is used for molecular quantitation and intermolecular interaction analysis. Since FPs are usually small soluble proteins, they easily cross the membranes if cell integrity is disrupted, resulting in FP signal attenuation/loss. Specimen prexation to preserve FP localization within cells/tissues is therefore useful. However, specic xatives can weaken or eliminate FP signals. We studied the effects of ve common xatives on FP uorescence intensity and biological functions to determine their suitability for FP signal and FRET eciency preservation in cells and tissues. FP (GFP, YFP, CFP and RFP)-expressing HEK293T cells with methanol, 95% ethanol, 4% PFA, acetone and glutaraldehyde, and brain tissue sections of EGFP-and tdTomato-labeled transgenic uorescent mice was xed with 4% PFA. The FP signals in HEK293T cells and brain tissue from transgenic uorescent mice were weakened or even eliminated after xation with these xatives. The xatives affected FP biological function, and the FP FRET eciency signicantly differed between prexation and postxation (all p<0.01). Thus, xatives impair FP uorescence to some extent, leading to attenuation/loss of signals or even biological functions. Fixatives should be applied carefully in FP-related experiments to avoid bias. decreased after PFA xation, consistent with the results of this study. The effects may be related to the mechanisms of the xatives; the specic mechanisms need further study. This experiment also used 4% PFA-xed brain tissue sections of transgenic mice labeled with EGFP and Ai4 transgenic mice labeled with tdTomato to explore the effect of xatives on FP signals. The results showed that the uorescence signals of EGFP and tdTomato in transgenic uorescent mouse brain tissues decreased after xation with 4% PFA. (Gibco BRL). After 5 hours, the transfection mixture was replaced with 1 ml of culture medium. The cells were used 2 days after transient transfection. The uorescence of the cells was visible, and 70–80% of the cells exhibited bright uorescence under confocal laser scanning microscopy. These infected cells were then prepared for the experiments. Cells without plasmid or lentivirus infection were used as negative controls. All processes in this study involving plasmid or lentivirus were performed in a


Introduction
Fluorescent proteins (FPs) are commonly used reporter proteins in biology. They are often expressed in cells in the form of fusion proteins [1][2] . FPs, which are frequently used as probes for coding genes and can highlight proteins and cells of interest, are used in a wide range of applications, such as uorescence imaging, uorescence resonance energy transfer (FRET) techniques, and ow cytometry [3][4][5][6] . In addition, uorescence in transgenic animals with the ability to spontaneously uoresce upon excitation with light of the appropriate wavelength, such as transgenic uorescent mice, provides scientists with a relatively simple and nontoxic biological tool and a tracing tool for cell therapy [7][8] . Before uorescence imaging or FRET, it is generally necessary to x cells or tissues with speci c xatives in order to maintain antigenicity and to prevent the FPs or cells from shifting during the experiment and affecting the downstream experimental results 9 . However, improper selection of xatives for cells or tissues labeled with reporter FPs may lead to changes in the spatial conformations of FPs 10 . Such changes are problematic for FP detection, accurate localization and related microstructural analyses. Therefore, it is of practical signi cance to explore the in uences of xatives on FP signals.
FRET with FPs is a powerful method for the detection of protein-protein interactions, enzyme activity and small molecules in the intracellular milieu [11][12] , and it can also be used to determine the distance between two different uorescent groups in real time [13][14][15][16] . Accurate measurement of the increased uorescence intensity of the acceptor group and the reduced uorescence intensity of the donor group enables accurate quanti cation of the e ciency of FRET 17 . Living cells usually need to be xed before FRET to preserve the cell structure and to prevent translocation of FPs and cells. In addition, for studies on transgenic uorescent animals labeled with uorescent reporter genes, which can be used to directly monitor cell activity and gene behavior in organisms with sensitive detection instruments, tissue xation can help optimize sample preservation and the FP signal.
However, improper selection and use of xatives may in uence FP signals, affecting FP biological functions. In the current study, we applied ve routinely used xatives to investigate the effects of xatives on the uorescence intensity and FRET e ciency of FPs in HEK293T cells and transgenic uorescent mice labeled with FPs and to further evaluate xatives that produce minimal loss of FP uorescence signals and biological functions in cells or tissues.

Results
HEK293T cells were infected with plasmids expressing GFP and RFP and with a lentivirus expressing YFP and CFP. Approximately 70-80% of cells revealed bright uorescence under a confocal laser scanning microscope after culture for 48 h. There was no obvious damage in the cells. Then, the cells expressing FPs were xed at room temperature, and the changes in the FP uorescence signals were detected under a confocal laser scanning microscope in real time (Fig. 1b, Fig. 2).
The tested xatives are used in research and have different stabilization mechanisms; thus, they have different effects on FP signals. Cells were post xed with methanol, 95% ethanol, 4% paraformaldehyde (PFA), acetone and glutaraldehyde at room temperature. Over 10 min, the YFP signal weakened rapidly, and with time, the YFP signal weakened further. Among cells xed with methanol and 95% ethanol, the YFP signal weakened signi cantly. Surprisingly, almost no YFP uorescence signal was detectable after xation with 4% PFA, while relatively strong uorescence signals were preserved after xation with acetone and glutaraldehyde. Post xation with methanol and 4% PFA signi cantly decreased CFP uorescence, while a strong signal remained after post xation with 95% ethanol, acetone and glutaraldehyde. Both CFP uorescence and YFP uorescence were weakened after xation with the ve kinds of xatives. The percentages of CFP and YFP uorescence preservation were 14.71% and 19.27%, respectively, after methanol xation. The CFP and YFP uorescence intensities after xation with acetone were better than those after xation with methanol (44.24% and 62.64%), respectively, and approximately 1/3 of the CFP and YFP uorescence intensity remained after glutaraldehyde xation (Fig. 1, Table 1).
In the rst 5 min of methanol, 95% ethanol and acetone xation, the GFP and RFP uorescence was reduced to less than 20% of the initial uorescence, while 4% PFA xation preserved more than 75% of the uorescence. However, as the xation time increased, the uorescence signals of GFP and RFP decreased gradually. When the xation time reached 15 min, the uorescence intensities of GFP and RFP were 54.56% and 56.29% for the 4% PFA xation group, respectively; the intensities were greater than 35% for the glutaraldehyde xation group (Fig. 2, Table 1). To test whether xatives can also affect the biological functions of FPs by exerting unfavorable effects on FP signals, we performed FRET. As CFP and YFP are commonly used together as a pair, CFP and YFP were selected for the FRET experiments. Twenty-ve independent CFP-and YFP-co-transfected cells were randomly selected for FRET from among living and xed cells. There were signi cant differences in FRET e ciency between pre xation and post xation with alcohol (P = 0.005), 95% ethanol (P = 0.001), acetone (P = 0.001), 4% PFA (P < 0.01) and glutaraldehyde (P < 0.01). The FRET e ciency increased when 95% ethanol was used; in contrast, it decreased when methanol, acetone, 4% PFA and glutaraldehyde were used. Among the xatives, the FRET e ciency of CFP and YFP decreased most obviously after xation with 4% PFA (Fig. 3).
Among all of the xatives, 4% PFA best preserved the uorescence signals of GFP and RFP; thus, 4% PFA was thus used to x brain tissue sections of transgenic uorescent mice labelled with EGFP and tdTomato. Figure 4 shows that the post xation uorescence intensities of brain sections were weaker than the pre xation intensities; however, there was no signi cant difference between pre-and post xation (p > 0.05).
Taken together, these results show that xatives unfavorably affect FP uorescence in HEK293T cells and tissues of transgenic uorescent mice to a certain extent and can thus lead to biased results regarding biological functions in analyses such as FRET.

Discussion
Methanol, 95% ethanol, 4% PFA, acetone and glutaraldehyde are routinely used as xatives in biomedical experiments. In many studies, cells or tissues need to be xed for subsequent microscopic observation, probe hybridization, or immunochemical staining. Before the relevant experiments are performed, the corresponding xatives should be selected according to the speci c cells, tissues and experimental purposes. FPs are applied as reporter genes and are widely used in gene transfection and expression studies and for in vivo uorescence imaging of transgenic animals to observe the movement and localization of target proteins in cells. In addition, transgenic uorescent animal models provide powerful tools for molecular imaging studies and studies on tumor biology, immune-tumor cell interaction, tumor angiogenesis, and stem cell differentiation and localization. Although FPs are widely applied in biomedical research, few studies have reported changes in the uorescence intensity of FPs during cell or tissue xation. Only sporadic information in relevant reports is available to researchers as guidance. Unfortunately, the descriptions of cell or tissue xation methods in most studies are very simple, making it di cult for other experimenters to accurately determine the methods that were used. Therefore, the effects of some routinely used xatives on the uorescence intensity of FPs need to be elucidated.
Fixatives commonly applied in general molecular biology experiments are divided into two categories: organic solvents and crosslinkers 18 . These xatives are used for research and diagnostics and act through different stabilization mechanisms. The main organic solvent xatives are methanol, ethanol and acetone. Methanol is a strong dehydrating agent that can immobilize cells in a certain con guration and increase the surface area of cell structures 19 . Ethanol is a small-molecule nonpolar permeabilizing substance that can penetrate cells quickly, x the ne structures of cells, dehydrate tissues/cells, and denature proteins via precipitation to x cells 20 . Acetone is a xative with permeabilization, strong dehydration, and protein precipitation abilities 19 . Paraformaldehyde and glutaraldehyde are routinely used crosslinking agents. The aldehyde group in the paraformaldehyde molecule binds with the amino groups of proteins to form carboxymethyl groups, thus forming intermolecular crosslinks that affect protein con gurations and x the proteins 21 . Glutaraldehyde is a bifunctional crosslinking agent 22 that has two reactive aldehyde groups that can undergo acylation with amino groups, mercapto groups and other groups, enabling it to crosslink with proteins. The changes in the molecular structure and conformation of cells induced by xation may change the uorescence intensity and protein interactions of FPs. Therefore, selection of an appropriate xative is crucial for accurate detection of the protein of interest and its biological function.
Previous studies have reported that the GFP signal in GFP-expressing cells can be successfully maintained using PFA [23][24] . Our study showed that FP uorescence decreased after xation with 4% PFA at room temperature and that CFP and exhibited the most signi cant uorescence loss, while strong signals were preserved for GFP and RFP. According to some studies, FP signals decrease or disappear after xation with 4% PFA; these ndings may be related to deviation from a neutral pH during the preparation of 4% PFA. The pH value of 4% PFA in this experiment was close to neutral, so interference of the pH value with the experimental results was excluded. In another study with 4°C PFA, the GFP signal was barely detectable, indicating that crosslinking of proteins was too slow to retain GFP; in contrast, more con ned and intense GFP signals were obtained in sections post xed with 25°C PFA 25 . Interestingly, another xative called glyoxal has recently been proposed to be a relevant alternative to PFA, but whether glyoxal works as a post xative to effectively preserve FP signals remains to be investigated. Few studies have reported the effect of glutaraldehyde on FP signals. In this study, glutaraldehyde also decreased FP uorescence; approximately 1/3 of the uorescence was preserved.
Previous studies have shown that acetone can preserve GFP signals as well. The results of our study showed that acetone weakened the FP signals, among which the GFP and RFP signals decreased most obviously. However, the CFP and YFP signals remained strong. The different mechanisms of acetone and paraformaldehyde xation in cells provide choices for experimental studies using different FPs as gene probes.
Some studies have reported that methanol can also be applied to maintain GFP uorescence 26 , but our study showed that methanol may lead to weakening or even disappearance of GFP uorescence 10 . In fact, the uorescence signals of all four FPs decreased signi cantly after xation with methanol, indicating that methanol could not effectively preserve the FP signals. Therefore, methanol is not recommended for xation of cells expressing FPs.
Ethanol-mediated cell xation has been found to cause GFP uorescence to disappear quickly 5 . In agreement with this nding, our study revealed that the uorescence intensity of the four FPs was weakened rapidly after xation with 95% ethanol. Among the signals, the RFP signal was lost most obviously, while some of the uorescence intensity of CFP was preserved. Some studies have reported that methanol and ethanol alone or mixed with acetone and formaldehyde can be used as xatives. However, unless otherwise required for speci c experiments, for the types of FPs applied in our study, acetone is recommended for xation to maintain CFP and YFP uorescence, while 4% PFA is recommended for xation to preserve GFP and YFP uorescence.
Another question is whether increases or decreases in FP uorescence during xation will affect the results of related experiments based on the uorescence intensity of FPs. To answer this question, we selected two interacting proteins, ADORA2a and DRD2, for co-expression with YFP and CFP, respectively. We co-transfected them into HEK293T cells with lentiviruses and then detected the FRET e ciency. The optimum distance of FRET detection is 1-10 nm; when the spatial displacement of FP molecules is too large, it affects the detection e ciency of FRET, especially for cells with strong molecular motion of FPs. In such cases, it is necessary to x the cells in order to retain the interactions between protein molecules in a particular cell state and to perform the FRET experiment smoothly.
In our study, the xatives had signi cant negative impacts on FRET e ciency (all P < 0.01). The FRET e ciency after 95% ethanol xation was higher than that before xation, while the e ciencies after methanol, 4% PFA, acetone and glutaraldehyde xation were all lower than those before xation. The FRET e ciency decreased most obviously after xation with 4% PFA. Previous studies have shown no signi cant differences in FRET e ciency between un xed living cells and living cells xed with 4% PFA 27 . Chu Y W et al. 24 showed that FRET e ciency decreased after PFA xation, consistent with the results of this study. The effects may be related to the mechanisms of the xatives; the speci c mechanisms need further study.
This experiment also used 4% PFA-xed brain tissue sections of transgenic mice labeled with EGFP and Ai4 transgenic mice labeled with tdTomato to explore the effect of xatives on FP signals. The results showed that the uorescence signals of EGFP and tdTomato in transgenic uorescent mouse brain tissues decreased after xation with 4% PFA.
According to the results of our study, the effects of xatives on FP signals in cultured cells and transgenic uorescent mouse tissues are unfavorable to some extent. Our ndings also con rmed that these effects can cause bias in FRET e ciency. Whether the effects also cause errors in the results of other experiments based on the uorescence intensities of FPs, such as ow cytometry or immunohistochemistry, remains to be examined.
Taken together, our ndings reveal that xatives indeed have certain unfavorable effects on FP signals in cells and tissues that lead to decreases in or even loss of their biological functions. Therefore, xatives should be applied carefully in FP-related experiments to avoid bias in the experimental results.

Methods
Cell culture and transfection.
HEK293T cells, which was purchased from ATCC, were cultivated on 6-well plate (Nunc) at 37°C in 95% air and 5% CO 2 . The medium was DMEM (Gibco, Invitrogen Corporation, NY, United States) containing 10% fetal bovine serum (FBS, Gibco), 100 units/ml penicillin, and 100 µg/ml streptomycin (Gibco). The cells were transfected 1 day after cultivating, at which time they were 80% con uent. The cells on a 6-well plate were transfected with a mixture of 1 mg of plasmid DNA encoding four kinds of FP-fusion constructs and 12 mg of Lipofectamine reagent (Gibco BRL, Gaithersburg, MD) in a 0.2-ml vol of Opti-MEM (Gibco BRL). After 5 hours, the transfection mixture was replaced with 1 ml of culture medium. The cells were used 2 days after transient transfection. The uorescence of the cells was visible, and 70-80% of the cells exhibited bright uorescence under confocal laser scanning microscopy. These infected cells were then prepared for the experiments. Cells without plasmid or lentivirus infection were used as negative controls. All processes in this study involving plasmid or lentivirus were performed in a biosafety level 3 (BSL-3) facility.
Cell processing.
The cell culture medium was removed, and the cells were washed with phosphate buffer saline (PBS) 3 times. The uorescence of living cells was observed under a confocal laser scanning microscope. The method of acceptor photobleaching (acceptor bleaching, AB) was applied to detect the uorescence resonance energy transfer (FRET) e ciency between CFP and YFP. The cells were rst xed at room temperature for 10 min 5 using ve kinds of xatives and then diluted with 4 ml sterile water. The changes in FP signals were detected by confocal laser scanning microscopy in real time. The monitoring time was 15 min in total. After 15 min, the xatives were discarded, and the cells were gently and carefully washed 4 times with PBS. The FRET experiment was then conducted again as post-xative results.
Ethics statements.
EGFP-and tdTomato-expressing transgenic mice, which were purchased from the Jackson Laboratory, were used for the experiment. For studies on transgenic mice, we con rm that all methods were carried out following relevant guidelines and regulations. Furthermore, we con rmed that the study was carried out in compliance with the ARRIVE guidelines. All mice in the study were maintained and used according to protocols approved by the Institutional Animal Care and Use Committee of the Peking University Health Science Center (Ethics number: LA2020310). The mice housed in Transgenic Biosafety BSL-3 laboratories of the Peking University Health Science Center Laboratory Animal Application and Research Center. All processes of the animal experiment were in line with recommendations for the care and use of laboratory animals.
Tissue processing.
The EGFP and tdTomato expressing transgenic uorescent mice were sacri ced according to normal pathological procedures. Brie y, mice were sacri ced by intraperitoneal injection of pentobarbital sodium (150mg/kg). Mouse brains were dissected and snap-frozen sliced at a 14um thickness, and then placed brain sections under a confocal laser scanning microscope to observe the uorescence intensities. After recording the data, the brain sections were xed with 4% paraformaldehyde (PFA) for 12 hours. After 3 washes with PBS, the sections were observed under a confocal laser scanning microscope to record the uorescence intensities again.
Cells and tissues were visualized using a confocal laser scanning microscope (Leitz Co.; with Technical Instruments coaxialconfocal attachment). The cells were viewed using a 100× oil immersion objective, and the tissues were viewed using a 40× objective.
Paired-sample t tests were used to analyze the two groups of data. P < 0.05 was considered to indicate a signi cant difference.   Effects of ve kinds of xatives on the FRET e ciency of CFP and YFP before and after xation. **P < 0.01 was considered to indicate a signi cant difference.