The Density of GM1-Enriched Lipid Rafts Correlates

Inversely with the Efficiency of Transfection Mediated

by Cationic Liposomes

Tamas Kovacs,1 Andrea Karasz,1 Janos Szollo†si,1,2 Peter Nagy1*

� AbstractAlthough cationic liposome-mediated transfection has become a standard procedure,the mechanistic details of the process are unknown. It has been suggested that endocyticuptake of lipoplexes is efficient, and transfectability is largely determined by later steps.In this article, we stained GM1-enriched membrane microdomains, a subclass of lipidrafts, with subunit B of cholera toxin and correlated transfection efficiency with theirdensity by quantitatively evaluating microscopic images. We found a strong anticorrela-tion between the density of GM1-enriched membrane microdomains and the efficacyof transfection monitored by measuring the expression level of GFP in different celllines transfected by lipofection using two different transfection agents. These findingsimply that GM1-enriched membrane microdomains interfere with the process of lipo-fection. The blocked step must be endocytosis since the accumulation of fluorescentlylabeled plasmids was lower in cells with high content of GM1-enriched membranemicrodomains. Such a correlation was not observed in cells transfected by electropora-tion. By comparing the efficiency of lipofection in several cell lines we found that thosewith a high density of GM1-enriched membrane microdomains were the most resistantto transfection. We conclude that the inhibition of lipofection by GM1-enriched mem-brane microdomains is a general rule, and that endocytosis of lipoplexes can be ratelimiting in cells with high density of GM1-enriched membrane rafts. ' 2009 International

Society for Advancement of Cytometry

� Key termslipid raft; lipoplex; transfection efficiency; CTX-B; GM1 ganglioside

TRANSFECTION of cells with DNA or siRNA has become a routine procedure in

molecular biology. Although viral transfer of genes (transduction) yields a high frac-

tion of transfected cells with high protein expression levels (1), its acceptance in

experimental and clinical applications is hampered by safety concerns (2). Nonviral

gene transfer can be broadly divided into two types: (i) physical methods including

electroporation and microinjection; (ii) chemical approaches using cationic lipo-

somes, dendrimers, or calcium phosphate (3). Although probably there is not a single

molecular biology lab in the world where cationic liposome-mediated transfection is

not used, the mechanism of DNA transfer is unclear, and transfection protocols are

empirical. Currently used second generation liposome formulations contain a mix-

ture of a cationic lipid and a helper lipid (4). Plasmid DNA is large and charged;

therefore, it encounters barriers during the transfection process. Electrostatic interac-

tions between cationic lipids and the negatively charged nucleic acid lead to the

formation of DNA–lipid complexes (lipoplexes) in which DNA is neutralized, con-

densed, and protected from degradation (5). It is widely accepted that lipoplexes are

endocytosed (6), and that productive endocytosis leading to gene expression is gener-

ally achieved by the clathrin-mediated pathway (7), although lipid raft/caveolae-

dependent endocytosis has also been shown to play a role under certain circumstances,

1Department of Biophysics and CellBiology, University of Debrecen,Debrecen 4012, Hungary2Cell Biophysical Workgroup of theHungarian Academy of Sciences,Research Center for MolecularMedicine, University of Debrecen,Debrecen 4012, Hungary

Received 10 April 2009; RevisionReceived 13 May 2009; Accepted 18 May2009

Additional Supporting Information may befound in the online version of this article.

Grant sponsor: Hungarian Scientific Re-search Fund, Grant numbers: OTKA72677, 68763; Grant sponsor: EuropeanCommission; Grant numbers: LSHB-CT-2004-503467, LSHC-CT-2005-018914,MCRTM-CT-2006-0359462.

*Correspondence to: Peter Nagy,Department of Biophysics and CellBiology, University of Debrecen,Nagyerdei krt 98, Debrecen 4012,Hungary

Email: nagyp@dote.hu

Published online 12 June 2009 in WileyInterScience (www.interscience.wiley.com)

DOI: 10.1002/cyto.a.20756

© 2009 International Society forAdvancement of Cytometry

Original Article

Cytometry Part A � 75A: 650�657, 2009

especially in gene transfer mediated by polyplexes (8,9). Since

endocytosis of lipoplexes is more efficient than later steps of

the transfection process, there are more cells taking up plas-

mid DNA than those expressing it (10). Rate limiting steps of

transfection are thought to include escape of the lipoplex from

endosomes, release of DNA from the lipoplex and nuclear

import (5,11,12). Helper lipids are assumed to play an instru-

mental role in the escape of the DNA from the endosomal

compartment by inducing lipid phase transition (4). In parti-

cular, the superiority of multicomponent lipoplexes has been

shown to be the consequence of their ability to induce rupture

of the endosomal membrane (13).

Lipid rafts are dynamic membrane microdomains with a

peculiar composition characterized by high cholesterol, sphin-

golipid, and ganglioside content (14). Since the entity defined

as raft depends on the method used for its detection, lipid rafts

prove to be difficult to investigate (15). Flow cytometric

(16–18) and complex microscopic techniques (19) have gained

importance in the investigation of lipid rafts due to their

potential for medium to high throughput assays or their ability

to reveal raft heterogeneity. Lipid rafts have been implicated in

organizing transmembrane signaling and membrane trafficking

(20), but their relationship to lipoplex-mediated transfection

has been barely investigated. It has been reported that in HeLa

cells raft-mediated internalization of polyethylenimine poly-

plexes was more efficient than the clathrin-mediated pathway

(8), but productive endocytosis of DNA-lipoplex complexes

followed the clathrin-mediated pathway (9). Since glycosyl-

phosphatidylinositol (GPI)-anchored proteins and ganglioside

GM1 accumulate in lipid rafts, both of them are used as speci-

fic raft markers (21,22). Lipid rafts are heterogeneous with

regard to their composition and function (23,24). Specifically,

GM1-enriched rafts show only partial overlap with microdo-

mains containing GPI-anchored proteins (25).

In this article, we used subunit B of cholera toxin (CTX-

B) to specifically label and quantitate the density of GM1-

enriched membrane microdomains, a subtype of lipid rafts.

We show that a high density of GM1-enriched membrane

microdomains inhibits transfection mediated by cationic lipo-

somes, and conclude that endocytosis is the rate limiting step

in cationic liposome-mediated transfection in cells with a high

density of GM1-enriched membrane microdomains. This

principle can be used to design more powerful transfection

agents rationally.

MATERIALS AND METHODS

Cells

The breast cancer cell line JIMT-1, available from the

German Collection of Microorganisms and Cell Cultures

(www.dsmz.de), was grown in F-12/DMEM (1:1) supplemented

with 20% FCS, 60 units/L insulin and antibiotics (26). The

human breast cancer cell line SKBR-3, the human cervix adeno-

carcinoma cell line HeLa, the human epithelial carcinoma cell

line A431, the mouse fibroblast cell line NIH/3T3 and the Chi-

nese hamster ovary cell line CHO were obtained from the

American Type Culture Collection (Rockville, MD) and grown

according to their specifications. The immortalized human

keratinocyte cell line HaCaTwas obtained from the Department

of Physiology, University of Debrecen and cultured in DMEM

supplemented with 10% FCS and antibiotics. For microscopic

experiments, cells were cultured on Lab-Tek II chambered cov-

erglass (Nalge Nunc International, Rochester, NY).

Transfection and Labeling of Cells, Plasmids

Cells grown on 2-well chambered coverglass were trans-

fected with Lipofectamine2000 (Invitrogen, Carlsbad, CA)

using 1.5 lg DNA/well and a lipid to DNA ratio of 2:1 (v/w).

Transfection with Effectene (Qiagen, Valencia, CA) was carried

out at an Effectene:Enhancer:DNA ratio of 25:8:1 (v/v/w). The

transfection protocols were otherwise according to the manu-

facturers’ specifications. Electroporation was performed by the

nucleofector device of Amaxa (Cologne, Germany) using

solution V and protocol T-20.

The GFP plasmid pmaxGFP was purchased from Amaxa

(Cologne, Germany). The GFP-GPI plasmid was a kind gift

from Jennifer Lippincott-Schwartz (NIH, Bethesda, MD). An

irrelevant plasmid (pSUPER from Oligoengine, Seattle, WA)

was fluorescently labeled by performing nick translation in the

presence of fluorescein-dUTP using the nick translation kit of

Roche (Cat. No. 10,976,776,001) according to the protocol

provided by the manufacturer. GM1-enriched membrane rafts

were labeled by incubating cells in the presence of 8 lg/ml

AlexaFluor647-tagged subunit B of cholera toxin (CTX-B,

Molecular Probes-Invitrogen, Eugene, OR) for 30 min on ice

to prevent internalization of CTX-B. Afterwards, cells were

washed twice in PBS and fixed in 1% formaldehyde. For label-

ing the light chain of clathrin cells were fixed in 3.7% formal-

dehyde for 30 min, stained with Ab-1 (clone CON.1, Dianova,

Hamburg, Germany) in PBS containing 0.1% BSA and 0.1%

TX-100 followed by labeling with an anti-mouse secondary

antibody tagged with AlexaFluor488.

Transferrin Uptake Experiment

Cells were starved in medium199 containing 0.1% FCS

for 1 h, then incubated in the same type of medium contain-

ing 10 lg/ml AlexaFluor488-labeled transferrin (Molecular

Probes-Invitrogen) for 60 min at 378C followed by staining

with CTX-B as described earlier.

Confocal Microscopy and Image Analysis

Image acquisition was performed using an LSM510 confo-

cal laser scanning microscope (Carl Zeiss AG, Gottingen, Ger-

many). GFP was excited with the 488 nm line of an argon ion

laser and its fluorescence was detected between 505 and 530

nm. AlexaFluor647 was excited with the 633 nm line of a red

He-Ne laser, and its emission was measured over 650 nm. Fluo-

rescence images were taken as 1-lm optical sections using a

403 (NA5 1.3) or 633 (NA5 1.4) oil immersion objective.

Image processing was carried out with the DipImage

toolbox (Delft University of Technology, Delft, The Nether-

lands) under Matlab (Mathworks, Natick, MA). Segmentation

of images into membrane and nonmembrane pixels was car-

ried out with the manually seeded watershed algorithm using

ORIGINAL ARTICLE

Cytometry Part A � 75A: 650�657, 2009 651

a custom-written Matlab program as described previously

(27,28). Cells were analyzed on a cell-by-cell basis, and single

cells were identified by the flood-fill operation carried out on

the image segmented by the watershed algorithm. Each dot in

the presented dot plots represents the mean fluorescence

intensity of a single cell, and a dot plot contains data of 500–

1,000 cells. The fluorescence intensities of GFP, GFP-GPI,

clathrin, and transferrin were measured in the whole cell,

whereas that of CTX-B was calculated in the membrane. The

membrane mask around each cell was determined by taking

the boundary around the object, i.e., the cell, thickened by 1

pixel both inwardly and outwardly. The background-corrected

mean fluorescence intensities of single cells calculated for the

whole cell (GFP, GFP-GPI, clathrin, transferrin) or for the

membrane (CTX-B) are presented. Background correction

was carried out by subtracting the mean fluorescence intensity

of a cell-free area from each pixel in the image.

RESULTS

The Efficiency of Lipofection of JIMT-1 Cells Inversely

Correlates with the Density of GM1-Enriched

Membrane Microdomains

Since the lipid raft content of single cells has not been

related to the efficiency of transfection with cationic lipo-

somes, we examined whether the two parameters are corre-

lated. In the first experiments, JIMT-1 cells were transfected

with GFP-GPI plasmid using Lipofectamine2000, and GM1-

enriched membrane rafts were stained by fluorescently labeled

subunit B of cholera toxin (CTX-B) 48 h after transfection.

Since GPI-anchored proteins are known to accumulate in lipid

rafts, the GFP and CTX-B signals were expected to show

strong positive correlation. Surprisingly, the two signals were

strongly anticorrelated in the majority of cells, while a minor

subpopulation showed the expected positive correlation (Fig.

1A). The conclusion of the quantitative analysis is visually

supported by the confocal microscopic image in which most

of the cells are either green (GFP-GPI) or red (CTX-B), and a

couple of them appearing in yellow show colocalization of

GFP-GPI and CTX-B (Supporting Information Fig. S1).

To exclude the possibility that the observed anticorrela-

tion is the consequence of lipid raft heterogeneity (24) or

GFP-GPI-induced changes in lipid raft structure, cells were

transfected with GFP plasmid. Not a single cell displayed the

yellow color characteristic of colocalization between GFP and

CTX-B (Fig. 2), and an even stronger anticorrelation was

revealed by the quantitative analysis of single cells (Fig. 1B).

Cells with a high density of GM1-enriched membrane micro-

domains were never transfected, and cells showing high pro-

duction of the transfected protein were always very dim in the

CTX-B channel. The negative correlation between the density

of GM1-enriched rafts and transfection efficiency was also

revealed by plotting the relationship between mean GFP fluo-

rescence as a function of raft density (CTX-B fluorescence,

Fig. 3A). To rule out the possibility that transfection by cati-

onic liposomes influences the binding of CTX-B we compared

the mean fluorescence intensities of control and GFP-trans-

fected cells and observed no significant difference (Fig. 4). We

tentatively concluded that a high density of GM1-enriched

membrane microdomains somehow interferes with the

process of lipofection.

The Endocytosis of a Fluorescently Labeled Plasmid is

Correlated Inversely with the Density of GM1-Enriched

Membrane Microdomains

Although lipoplex uptake is considered to be more

efficient than later steps of transfection (5,29), the assumption

that a high density of GM1-enriched membrane microdo-

mains interferes with endocytic uptake of DNA seemed

reasonable. Therefore, we transfected JIMT-1 cells with an

irrelevant plasmid (pSUPER) fluorescently labeled by nick

translation in the presence of fluorescein-dUTP, and stained

them with AlexaFluor647-CTX-B 8 h after transfection.

Confocal microscopy revealed a remarkable anticorrelation

between the plasmid content of single cells and their

Figure 1. The density of GM1-enriched membrane microdomains inversely correlates with lipofection efficiency. JIMT-1 cells were trans-

fected with GFP-GPI (A), GFP (B) or an irrelevant fluorescently labeled plasmid (C) using Lipofectamine2000 followed by staining with CTX-

B 2 days after transfection. The fluorescence intensity of the expressed protein (A,B) or that of the plasmid (C) is plotted as a function of

CTX-B fluorescence intensity. Each dot in the figure represents the mean fluorescence intensity of single cells measured in the respective

fluorescence channel of the confocal microscope.

ORIGINAL ARTICLE

652 Lipid Rafts Inhibit Transfection

GM1-enriched raft density (Figs. 1C and 5). We concluded

that a high density of GM1-enriched membrane microdo-

mains does not allow the efficient endocytic uptake of DNA-

liposome complexes.

The Inhibitory Role of GM1-Enriched Membrane

Microdomains is a General Property of Lipofection

To check whether the observed anticorrelation between

GM1-enriched raft density and transfectability by lipoplexes is

limited to the transfection of JIMT-1 cells with Lipofecta-

mine2000 we extended the investigation to a different cell line

and to a different transfection agent. Transfection of SKBR-3

cells with GFP using Lipofectamine2000 showed anticorrela-

tion between GFP intensity and the density of GM1-enriched

membrane microdomains, although the strength of the inverse

relationship was somewhat smaller than in JIMT-1 (Figs. 3B,

6A; Supporting Information Fig. S2). Transfection of JIMT-1

cells with GFP using a different lipid formulation (Effectene)

also resulted in negative correlation between transfection effi-

ciency and the density of GM1-enriched membrane microdo-

mains (Figs. 3B, 6B; Supporting Information Fig. S3). Assum-

ing that GM1-enriched membrane microdomains indeed

interfere with endocytic uptake of the lipoplex GFP expression

is expected to show no dependence on CTX-B staining after

electroporation. Analysis of JIMT-1 cells transfected by elec-

troporation with GFP revealed no correlation between the

density of GM1-enriched membrane microdomains and GFP

expression (Figs 3A, 6C, Supporting Information Fig. S4).

These results present convincing evidence that GM1-enriched

membrane rafts generally inhibit lipofection without influen-

cing the transfectability of cells by electroporation.

The Transfectability of Cell Lines is Predictable from

their Density of GM1-Enriched Membrane Rafts

Some cell lines are empirically known to be easily trans-

fectable, whereas others hardly lend themselves to transfection.

We selected seven cell lines including epithelial and fibroblast

lines, cancerous and noncancerous ones with known differ-

ences in their transfectability. All of them were transfected

with the same Lipofectamine2000-DNA complex and their

mean GFP expression levels were examined as a function of

the density of GM1-enriched rafts 48 h after transfection. The

data reveal a remarkably predictable tendency; cells with a

high density of GM1-enriched membrane microdomains are

difficult to transfect, while cells known and found to be easily

transfectable had low content of GM1-enriched membrane

rafts (Fig. 3C). We conclude that the level of GM1-enriched

membrane microdomains plays a fundamental role in deter-

mining the transfectability of cells with cationic liposomes.

Lack of Correlation Between the Expression Level of

Clathrin, Endocytic Uptake of Transferrin and the

Density of GM1-Enriched Membrane Microdomains

The inhibition of productive endocytosis of lipoplexes in

cells with a high density of GM1-enriched membrane micro-

domains could be explained by the low level or absence of

Figure 2. Staining with CTX-B shows strict anticorrelation with the expression of GFP. JIMT-1 cells were transfected with GFP using Lipo-

fectamine2000 and were stained with CTX-B 2 days after transfection. The fluorescence images of GFP (A), CTX-B (B), the phase contrast

image (C) and their overlay (D) reveal anticorrelation between the expression of GFP and CTX-B fluorescence intensities. Part E shows the

cells in color identified by the manually seeded watershed algorithm, whereas part F shows the membrane mask in red. GFP and CTX-B

fluorescence intensities were evaluated in the cell (E) and membrane (F) masks, respectively. The scale bar corresponds to 50 lm.

ORIGINAL ARTICLE

Cytometry Part A � 75A: 650�657, 2009 653

clathrin expression in these cells. To find evidence for or

against the aforementioned assumption we stained JIMT-1

cells with CTX-B and a monoclonal antibody against the light

chain of clathrin. The image (Fig. 7A) and its quantitative

evaluation (Fig. 4B) show a lack of correlation between the

density of GM1-enriched membrane microdomains and cla-

thrin expression level. Although clathrin expression did not

depend on the density of GM1-enriched lipid rafts, it was still

possible that clathrin-dependent endocytosis was inhibited by

a high density of GM1-enriched lipid microdomains. There-

fore, we incubated cells in the presence of fluorescently labeled

transferrin and stained them with CTX-B afterwards (Fig. 7B).

The uptake of transferrin depended negligibly on the density

of GM1-enriched membrane domains compared to that

observed for the uptake of fluorescent plasmids or the expres-

sion of transfected proteins (Fig. 4B).

DISCUSSION

Modification of gene expression by exogenous nucleic

acids (DNA, siRNA) is a widely used tool in research, and it

gains importance in clinical applications as well. Although

transfection of cells is most often carried out using cationic

liposomes, the efficiency of this approach in some cell types is

not as high as desirable. More importantly, factors influencing

Figure 4. CTX-B binding is independent of lipofection and clathrin expression. (A): JIMT-1 cells were transfected with GFP using Lipofecta-

mine2000, and nontransfected (thin solid line) and transfected cells (thick dashed line) were stained with AlexaFluor647-CTX-B 2 days after

transfection. For comparison nontransfected CHO cells were also stained with AlexaFluor647-CTX-B (thick solid line). CHO and JIMT-1 cells

were imaged with different photomultiplier settings, and the measured intensities of CHO cells were corrected for the higher gain. The his-

tograms show the distribution of the mean CTX-B intensity of single cells measured by confocal microscopy. (B): JIMT-1 cells were stained

with CTX-B and against the light chain of clathrin (thin solid line). In another experiment JIMT-1 cells were incubated in the presence of

fluorescently labeled transferrin for 1 h at 378C followed by staining with CTX-B (thick dashed line). The fluorescence intensity of singlecells was calculated and the range of CTX-B fluorescence intensity was divided into bins. The average clathrin or transferrin fluorescence

intensity in these bins was calculated and plotted against CTX-B intensity.

Figure 3. High density of GM1-enriched membrane microdomains in single cells and cell lines predicts low lipofection efficiency. (A,B):

JIMT-1 cells were transfected with GFP using Lipofectamine2000 (A, filled circle), electroporation (A, open triangle) or Effectene (B, open trian-

gle). SKBR-3 cells were transfected with GFP using Lipofectamine2000 (B, filled circle). Cells were stained with CTX-B 2 days after transfection,

and the cells were analyzed by confocal microscopy. The range of fluorescence intensities in the CTX-B channel was divided into 20 bins, and

the mean GFP fluorescence intensity was calculated separately for each bin. (C): Seven different cell lines were transfected with GFP using

Lipofectamine2000. The same lipid-DNA mixture was used for each cell line. Cells were stained with fluorescent CTX-B 2 days after transfec-

tion, and the fluorescence intensity of GFP and CTX-B were measured by confocal microscopy. The mean GFP intensity of the cell lines is

plotted against their average CTX-B intensities. The CTX-B intensity histograms of two of the cell lines (CHO, JIMT-1) are shown in Figure 4A.

ORIGINAL ARTICLE

654 Lipid Rafts Inhibit Transfection

distinct steps and the overall efficiency of transfection are

largely unknown making the procedure unpredictable or em-

pirical at best. Although gene transfer mediated by viral trans-

duction is more efficient, there are concerns about its safety

(2). Therefore, there is an urgent need for better gene transfer

technologies. A better understanding of the mechanisms of

cationic liposome-mediated transfection can lead to signifi-

cant improvements in the efficiency of this approach.

In this article, we observed a strong negative correlation

between the density of GM1-enriched membrane microdo-

mains and transfectability of cells by two commercial cationic

liposome formulations when the efficiency of transfection was

evaluated either at the level of lipoplex endocytosis or gene

transcription, i.e., protein expression. These findings imply

that a high density of GM1-enriched rafts blocks endocytosis

of the DNA-lipid complex. It has to be noted that caution

should be exercised in the generalization of the observed

inhibitory role of a high density of GM1-enriched rafts on

lipoplex endocytosis, since only two different commercial cati-

onic lipid mixtures were tested. We believe that it is still

Figure 5. Uptake of a fluorescent plasmid inversely correlates with the density of GM1-enriched membrane microdomains. JIMT-1 cells

were transfected using Lipofectamine2000 with an irrelevant plasmid (pSUPER) fluorescently labeled by nick translation, and they were an-

alyzed by confocal microscopy 8 h after transfection. (A) and (B) show the fluorescence of the plasmid and CTX-B, respectively, while (C) is

the phase contrast image. The overlay of the fluorescence and phase contrast images is shown in (D). (E) and (F) show the cell and mem-

brane masks, respectively. The scale bar corresponds to 30 lm.

Figure 6. The inverse correlation between transfection efficiency and the density of GM1-enriched membrane microdomains is a property

of lipofection. SKBR-3 cells were transfected with GFP using Lipofectamine2000 (A). JIMT-1 cells were also transfected with GFP using

Effectene (B) or by electroporation (C). Cells were stained by CTX-B 2 days after transfection and the cells were analyzed by confocal

microscopy. The mean intensity of GFP fluorescence is plotted against CTX-B fluorescence intensity on a cell-by-cell basis.

ORIGINAL ARTICLE

Cytometry Part A � 75A: 650�657, 2009 655

plausible to assume that the endocytosis of DNA complexed

with cationic liposomes is generally inhibited by a high density

of GM1-enriched lipid rafts, since the overall endocytic and

post-endocytic route of different lipoplex formulations is

known to be very similar (6).

The inhibition of cationic liposome-mediated transfec-

tion by GM1-enriched membrane microdomains seems to be

a general rule, since (i) it was observed in several cell lines; (ii)

it was reproduced with two transfection agents; (iii) it was not

observed with electroporation. The latter finding also supports

the conclusion that GM1-enriched membrane microdomains

inhibit the endocytic step of transfection, since in electropora-

tion DNA is taken up by cells through voltage-induced mem-

brane pores bypassing the endocytic machinery (30).

It is widely accepted that endocytosis of the lipoplex is

much more efficient than later steps of transfection, therefore

lipoplex internalization usually does not play a decisive role in

determining the outcome of transfection (5,29). However, the

results presented in the article show that endocytosis is the

limiting step (or one of the limiting steps) of transfection if

the density of GM1-enriched membrane rafts is high. We pro-

pose that the process of transfection is hindered in cells with a

high density of GM1-enriched lipid rafts at the level of endo-

cytosis. These cells take up a substantially lower amount of

lipoplex from which only insufficient quantities are able to get

out of endosomes and into nuclei. The inhibitory role of

GM1-enriched membrane microdomains is manifested at the

single cell level (i.e., within a population of cells) and when

different cell types are compared. Single cells and cell types

with a low density of GM1-enriched membrane microdomains

take up the lipoplex more efficiently, and later steps of the

transfection process limit the efficiency of transfection in

them.

We do not know the underlying principle behind the

inhibitory role of GM1-enriched membrane microdomains in

lipofection. Although it is usually accepted that uptake of

lipoplexes leading to gene expression proceeds via the

clathrin-dependent pathway (7), such a profound inhibition

of the process by a high density of GM1-enriched membrane

microdomains has not been observed. The fact that the

expression level of clathrin was independent of CTX-B stain-

ing and that transferrin endocytosis was only weakly

influenced by the density of GM1-enriched membrane micro-

domains argues that receptor-mediated endocytosis is not

generally inhibited by a high density of GM1-enriched rafts. It

is known that lipid rafts contain a high concentration of gang-

liosides each type of which contains one or more negatively

charged sialic acid. The high density of lipid rafts may change

the surface charge of cells in such a way which hinders the

interaction of the cell surface with lipoplexes which themselves

contain both positive and negative charges. It is worth men-

tioning that HeLa cells seemed to be an exception to the

inhibitory role of GM1-enriched membrane microdomains on

lipofection, since there was no correlation between the density

of GM1-enriched lipid rafts and transfection efficiency in this

cell line (our unpublished observation). The mechanism of

post-endocytic processing of internalized DNA complexes in

HeLa may be different from other cells, since DNA complexed

with polyethylenimine (PEI) polyplexes (branched or linear)

was taken up and expressed by HeLa cells by both the clathrin-

and raft/caveolae-mediated pathways, but the latter was in

general more efficient. In contrast, in other cell types (COS-7,

HUH-7) clathrin-dependent endocytosis was more relevant

for transfection by PEI (8). However, there is some contro-

versy since another study put forward that polyplex-mediated

transfection proceeds via the caveolae-dependent pathway in

general (9). However, lipoplex-mediated uptake of DNA was

found to be mediated by the clathrin-dependent pathway even

in HeLa cells (9). Although the evidence for the distinct post-

endocytic sorting of internalized DNA complexes by HeLa

cells is controversial, it can still be assumed that cell specific

parameters typical of HeLa cells may explain the lack of corre-

lation between transfection efficiency and GM1-enriched raft

density in these cells.

We have considered alternative explanations for the

observed anticorrelation between raft density, i.e., fluorescence

of AlexaFluor647-CTX-B, and transfection efficiency, i.e., the

fluorescence of fluorescein-labeled plasmid or that of GFP.

Fluorescence resonance energy transfer (FRET) between GFP

(or fluorescein) and AlexaFluor647 could account for the

observed anticorrelation, but the large separation between the

spectra of the dyes makes FRET very unlikely to happen. Alter-

natively, transfection could lead to such modifications in lipid

raft structure that would block the binding of CTX-B, i.e.,

instead of our assumption that GM1-enriched lipid rafts

inhibit transfection, the transfection process itself could alter

the binding of CTX-B to cells. However, we consider this unli-

kely since (i) the negative correlation was observed with differ-

ent transfection agents; (ii) expression vectors producing a raft

protein (GFP-GPI) or a cytoplasmic protein (GFP) lead to

similar results; (iii) the uptake of fluorescently labeled plasmid

which did not produce any protein was also inversely

Figure 7. Relationship between the density of GM1-enriched lipid

rafts, clathrin expression, and transferrin uptake. (A): JIMT-1 cells

were stained with CTX-B (red channel) followed by fixation, per-

meabilization and staining with a monoclonal antibody against

the light chain of clathrin (blue channel). The scale bar corre-

sponds to 15 lm. (B): JIMT-1 cells starved in the presence of 0.1%FCS in high iron medium (Medium199) were incubated in the pre-

sence of 10 lg/ml AlexaFluor488-transferrin (blue channel) for 60min at 378C followed by staining with AlexaFluor647-CTX-B (redchannel). The scale bar corresponds to 10 lm.

ORIGINAL ARTICLE

656 Lipid Rafts Inhibit Transfection

correlated with CTX-B intensity; (iv) the CTX-B intensity cor-

related inversely with the observed and known transfectability

of several cell lines; (v) the negative correlation between trans-

fection efficiency and CTX-B binding was not observed after

electroporation; (vi) lipofection did not lead to any change in

the binding of CTX-B. Therefore, we conclude that a high

density of GM1-enriched lipid rafts inhibits transfection

mediated by cationic liposomes.

GFP-GPI is used as a marker of lipid rafts (15,31). The

results presented in this article suggest that care should be

taken when interpreting these results, since GFP-GPI transfec-

tion does not label a subpopulation of cells in which the

density of GM1 gangliosides is very high. In addition to this

high-GM1, low-GFP-GPI subpopulation we found two more

subpopulations among these cells (Fig. 1A). There were cells

which had a low amount of GM1-enriched lipid rafts, there-

fore they expressed a high amount of GFP-GPI, since endocy-

tosis of the lipoplex was not inhibited in them. Although the

third subpopulation expressed a high amount of GM1-

enriched rafts, they got transfected, since the inhibition of

lipoplex endocytosis was not complete. These cells displayed

the expected positive correlation between the two lipid raft

markers.

In conclusion, the results presented in this article identify

high density of GM1-enriched lipid rafts as a strong negative

predictor of efficient transfection mediated by cationic lipo-

somes and suggest that productive endocytosis inhibited by

GM1-enriched membrane microdomains limits the efficiency

of transfection if the density of GM1-enriched lipid rafts is

high. The fact that cell lines with a high density of GM1-

enriched rafts exhibited low transfectability can be used to

select experimental model systems. In the future these princi-

ples can also be used to design better lipid formulations for in

vitro and in vivo gene transfer.

LITERATURE CITED

1. Cockrell AS, Kafri T. Gene delivery by lentivirus vectors. Mol Biotechnol 2007;36:184–204.

2. Raty JK, Lesch HP, Wirth T, Yla-Herttuala S. Improving safety of gene therapy. CurrDrug Saf 2008;3:46–53.

3. Gopalakrishnan B, Wolff J. siRNA and DNA transfer to cultured cells. Methods MolBiol 2009;480:31–52.

4. Wasungu L, Hoekstra D. Cationic lipids, lipoplexes and intracellular delivery of genes.J Control Release 2006;116:255–264.

5. Zhdanov RI, Podobed OV, Vlassov VV. Cationic lipid-DNA complexes-lipoplexes-forgene transfer and therapy. Bioelectrochemistry 2002;58:53–64.

6. Elouahabi A, Ruysschaert JM. Formation and intracellular trafficking of lipoplexesand polyplexes. Mol Ther 2005;11:336–347.

7. Zuhorn IS, Kalicharan R, Hoekstra D. Lipoplex-mediated transfection of mammaliancells occurs through the cholesterol-dependent clathrin-mediated pathway of endocy-tosis. J Biol Chem 2002;277:18021–18028.

8. von Gersdorff K, Sanders NN, Vandenbroucke R, De Smedt SC, Wagner E, Ogris M.The internalization route resulting in successful gene expression depends on both cellline and polyethylenimine polyplex type. Mol Ther 2006;14:745–753.

9. Rejman J, Bragonzi A, Conese M. Role of clathrin- and caveolae-mediated endocyto-sis in gene transfer mediated by lipo- and polyplexes. Mol Ther 2005;12:468–474.

10. Coonrod A, Li FQ, Horwitz M. On the mechanism of DNA transfection: Efficientgene transfer without viruses. Gene Ther 1997;4:1313–1321.

11. Zabner J, Fasbender AJ, Moninger T, Poellinger KA, Welsh MJ. Cellular and molecu-lar barriers to gene transfer by a cationic lipid. J Biol Chem 1995;270:18997–19007.

12. Salman H, Zbaida D, Rabin Y, Chatenay D, Elbaum M. Kinetics and mechanism ofDNA uptake into the cell nucleus. Proc Natl Acad Sci USA 2001;98:7247–7252.

13. Caracciolo G, Caminiti R, Digman MA, Gratton E, Sanchez S. Efficient escape fromendosomes determines the superior efficiency of multicomponent lipoplexes. J PhysChem B 2009;113:4995–4997.

14. Mayor S, Rao M. Rafts: Scale-dependent, active lipid organization at the cell surface.Traffic 2004;5:231–240.

15. Kiss E, Nagy P, Balogh A, Szollo†si J, Matko J. Cytometry of raft and caveolamembrane microdomains: From flow and imaging techniques to high throughputscreening assays. Cytometry Part A 2008;73A:599–614.

16. Morales-Garcia MG, Fournie JJ, Moreno-Altamirano MM, Rodriguez-Luna G, FloresRM, Sanchez-Garcia FJ. A flow-cytometry method for analyzing the composition ofmembrane rafts. Cytometry Part A 2008;73A:918–925.

17. Gombos I, Bacso Z, Detre C, Nagy H, Goda K, Andrasfalvy M, Szabo G, Matko J.Cholesterol sensitivity of detergent resistance: A rapid flow cytometric test for detect-ing constitutive or induced raft association of membrane proteins. Cytometry Part A2004;61A:117–126.

18. Wolf Z, Orso E, Werner T, Klunemann HH, Schmitz G. Monocyte cholesterol home-ostasis correlates with the presence of detergent resistant membrane microdomains.Cytometry Part A 2007;71A:486–494.

19. Gombos I, Steinbach G, Pomozi I, Balogh A, Vamosi G, Gansen A, Laszlo G, GarabG, Matko J. Some new faces of membrane microdomains: A complex confocal fluo-rescence, differential polarization, and FCS imaging study on live immune cells.Cytometry Part A 2008;73A:220–229.

20. Simons K, Toomre D. Lipid rafts and signal transduction. Nat Rev Mol Cell Biol2000;1:31–39.

21. Harder T, Scheiffele P, Verkade P, Simons K. Lipid domain structure of the plasmamembrane revealed by patching of membrane components. J Cell Biol 1998;141:929–942.

22. Zacharias DA, Violin JD, Newton AC, Tsien RY. Partitioning of lipid-modified mono-meric GFPs into membrane microdomains of live cells. Science 2002;296:913–916.

23. Wilson BS, Steinberg SL, Liederman K, Pfeiffer JR, Surviladze Z, Zhang J, SamelsonLE, Yang LH, Kotula PG, Oliver JM. Markers for detergent-resistant lipid rafts occupydistinct and dynamic domains in native membranes. Mol Biol Cell 2004;15:2580–2592.

24. Pike LJ. Lipid rafts: Heterogeneity on the high seas. Biochem J 2004;378:281–292.

25. Hofman EG, Ruonala MO, Bader AN, van den Heuvel D, Voortman J, Roovers RC,Verkleij AJ, Gerritsen HC, van Bergen En Henegouwen PM. EGF induces coalescenceof different lipid rafts. J Cell Sci 2008;121:2519–2528.

26. Tanner M, Kapanen AI, Junttila T, Raheem O, Grenman S, Elo J, Elenius K, Isola J.Characterization of a novel cell line established from a patient with Herceptin-resist-ant breast cancer. Mol Cancer Ther 2004;3:1585–1592.

27. Gonzalez RC, Woods RE, Eddins SL. Segmentation using the watershed algorithm.In: Gonzalez RC, Woods RE, Eddins SL, editors. Digital Image Processing UsingMatlab. Upper Saddle River: Pearson Prentice Hall; 2004. pp 417–425.

28. Palyi-Krekk Z, Barok M, Isola J, Tammi M, Szollo†si J, Nagy P. Hyaluronan-inducedmasking of ErbB2 and CD44-enhanced trastuzumab internalisation in trastuzumabresistant breast cancer. Eur J Cancer 2007;43:2423–2433.

29. Felgner PL, Ringold GM. Cationic liposome-mediated transfection. Nature1989;337:387–388.

30. Xie TD, Tsong TY. Study of mechanisms of electric field-induced DNA transfection.V. Effects of DNA topology on surface binding, cell uptake, expression, and integra-tion into host chromosomes of DNA in the mammalian cell. Biophys J1993;65:1684–1689.

31. Varma R, Mayor S. GPI-anchored proteins are organized in submicron domains atthe cell surface. Nature 1998;394:798–801.

ORIGINAL ARTICLE

Cytometry Part A � 75A: 650�657, 2009 657