ABSTRACT
DNA-calcium phosphate co-precipitates arise spontaneously in supersaturated solutions. Highly effective precipitates for transfection purposes, however, can be generated only in a very narrow range of physico-chemical conditions that control the initiation and growth of precipitate complexes. The concentrations of
calcium and phosphate are the main factors influencing characteristics of the
precipitate complex, but other parameters, such as temperature, DNA
concentration and reaction time are important as well. An example for this is
the finding that almost all of the soluble DNA in the reaction mix can be bound
into an insoluble complex with calcium phosphate in <1 min. Extending the reaction time to 20 min results in aggregation and/or
growth of particles and reduces the level of expression. With improved
protocols we gained better reproducibility and higher efficiencies both for
transient and for stable transfections. Up to 60% of cells stained positive for
[beta]
-gal and transient production of secreted proteins was improved 5- to 10-fold over results seen with transfections using standard procedures. Similar improvements in efficiency
(number of recombinant cell colonies) were observed with stable transfections,
using co-transfected marker plasmids for selection. Transient expression levels 2 days after DNA transfer and titers obtained from
stable cell lines, emerging weeks later, showed strong correlation.
Co-precipitates composed of `calcium phosphate' (hydroxyapatite) and purified DNA have been used for >20 years for the transfer to and
expression of genetic information in mammalian cells in culture (
1
). This technique has become one of the major methods for DNA transfer to
mammalian cells. A number of papers
(
2
-
6
) have addressed observed variability of DNA transfer and low efficiency and
they usually contain specific recommendations for `optimal' procedures.
However, these communications did not report on the complex relationships
between different components responsible for the creation of a supersaturated
status. We present here data that provide the basis for a better understanding,
at the physico-chemical level, of some of the most crucial aspects concerning the
formation of DNA containing precipitate complexes. We have applied gained insights in sets of
transfections both for transient and stable expression of recombinant proteins. We used two popular immortalized cell lines for these studies; human embryo kidney 293 cells (HEK-293) (
7
-
9
) and dihydrofolate reductase-minus (DHFR-) Chinese hamster ovary cells (CHO) (
10
-
13
). We were able to improve the efficiency of DNA transfer and expression in both
cell lines.
This paper also illustrates that a simple centrifugation-OD assay can be useful
in studying kinetics and reproducibility of precipitate formation. This test
provides information on the reliability of precipitate formation and gives
greater assurance for successful DNA transfer to cells.
Propagation of AmpR plasmids were done in the bacterial strain DH5a using 50 [mu]g/ml carbenicillin in LB medium.
DNA was isolated using the alkaline lysis method and was purified by double
equilibrium centrifugation in CsCl-ethidium bromide gradients as described by Sambrook
et al
. (
13
).
A solution of 100 [mu]l of 2.5 M CaCl
2
and the desired amount of plasmid DNA was diluted with 1/10 TE buffer (1 mM
Tris-HCl, 0.1 mM EDTA, pH 7.6) to a final volume of 1 ml. One volume of this 2* Ca/DNA solution was added quickly to an equal volume of 2* HEPES solution: 140 mM NaCl, 1.5 mM Na
2
HPO
4
, 50 mM HEPES, pH 7.05 at 23oC, or alternatively phosphate-free 2* HEPES buffer was prepared and supplemented with phosphate
from a 300 mM stock solution of pH 7.05 containing 195 mM Na
2
HPO
4
and 105 mM NaH
2
PO
4
. The two solutions were mixed quickly once, and added to the cell culture
medium after the time frame indicated.
Precipitate formation was confirmed and quantified by absorption at 320 nm against a blank lacking the phosphate or alternatively lacking both,
DNA and phosphate. The association (binding) of DNA with precipitate was
determined by OD measurement at 260 and 320 nm of a 250 [mu]l aliquot of the supernatant of the precipitation mixture after 30 s
centrifugation (16 000
g
) in an Eppendorf centrifuge. The supernatant was analyzed immediately after the centrifugation step. Precautions were undertaken to prevent heating up
of the rotor when multiple samples were analyzed. For precipitation experiments
at 0oC the rotor was cooled down on ice before it was used.
Cell culture was performed according to Doyle
et al.
(
21
). Both CHO cells and 293 cells were grown in a DMEM/F12 1:1 based medium,
supplemented with 2% fetal calf serum. CHO cells were maintained attached in 75
cm
2
flasks. Cells (293 HEK) were adapted to growth in suspension and grown in 250
ml spinner flasks. Both cell lines were subcultivated once or twice a week at
ratios between 1:10 and 1:100.
Cells from the exponential growth phase were seeded (1-4 * 10
5
cells/ml) into 12-well or 60 mm plates the day before the transfection was done. One hour
before the precipitate was added, the medium was replaced with fresh medium (pH
7.4). For each ml of medium, 100 [mu]l of precipitate was added. With a calcium concentration of 125 mM in the
precipitation mixture, the resulting final concentration of calcium in the cell
culture medium was ~12.5 mM. The cells were exposed to the precipitate for 2-6 h at 37oC at a pH of 7.3-7.6. For CHO cells, a glycerol shock was applied at this
point. The cells were exposed to 20% glycerol in PBS. After 1 min the glycerol
was removed by adding fresh medium, aspiration of the mixture and replacement
with fresh medium. For 293 cells the medium was replaced by fresh medium without applying a shock. The cells were then incubated for 1-6 days before the supernatant was harvested and analyzed by ELISA. Transfection
efficiency was determined by staining [beta]-galactosidase expressing cells with X-Gal after 24 h (
22
).
A 1:1 (w/w) mixture of linearized and purified plasmid containing DHFR as a selective marker or the TNK-tPA (
6
) expression cassette was precipitated as described above. The cells were grown
for 2 days after transfection in non-selective medium and the supernatant was harvested to measure transient
expression. The cells were trypsinized and seeded in 100 mm plates under different selective
pressure. Of these cells, 2% were seeded in 100 mm plates in medium lacking glycine, hypoxanthine and thymidine (GHT- medium), 20% in GHT- medium containing 30 nM methotrexate (MTX) and 40% at 100 nM MTX.
After 10 days, the product titers from pools of cells were assayed by ELISA. For each
level of selective pressure, titers seen from emerging pools of stable cells
were normalized against each other.
Subsequent to the mixing of the calcium/DNA solution with the HEPES/phosphate
solution a slight opacity appears within a few minutes, indicating that a
precipitate has been formed. With the intention of having the reaction
`complete', standard protocols suggest an incubation period of up to 20 (
4
) or even 30 min (
3
) at room temperature. To measure the rate at which the DNA is being incorporated
into or associating with the forming precipitate an assay was developed. This
test was based on a centrifugation step. The DNA concentration remaining in the
clarified solutions was assessed by determining the optical density at 260 nm. Surprisingly, at a pH of 7.05 and a DNA concentration of 25 [mu]g/ml, adsorption of DNA occurred within 30 s. In fact 30 s was the shortest
period for completion of this test. When either calcium or phosphate was
missing in the mixture, no DNA was found to be adsorbed (data not shown).
To determine the optimal incubation time for the formation of an efficient precipitate for transfections, the reaction mix was transferred to
cells at different time points upon initiation of the reaction. A [beta]-galactosidase ([beta]-gal) plasmid (
14
) was used to estimate the number of transfected cells (transfection efficiency). An expression vector for
human tissue plasminogen activator (TNK-tPA) (
16
,
16
) was used in a second transfection experiment.
Highest transfection efficiencies were observed with precipitation mixtures that had been incubated for short periods of time. In a representative
experiment ~60% of CHO cells and 38% of HEK-293 cells stained positive for [beta]-galactosidase in plates which had been exposed to 1 min precipitate complexes
(Fig.
1
A). The transfection efficiency decreased to 3-5% when precipitates were used after an incubation period of 20 min.
Using a 5 min reaction time gave intermediate results. Similar differences, here in yield of a
secreted protein, were observed with a transfected TNK-tPA vector (Fig.
1
B). Almost 2 [mu]g/ml for HEK-293 cells and 0.5 [mu]g/ml for CHO cells of TNK-tPA were detected by ELISA in the supernatant of cells
exposed to an `early' precipitate-complex (1 min). `Late' precipitates (20 min) gave only 10% of these
titers, confirming observations made by O'Mahoney
et al
. (
6
).
These differences in transfection efficiency or in the levels of secreted
recombinant protein could be correlated with the nature of the calcium
phosphate precipitate in plates (Fig.
2
). Precipitates added 1 min after mixing, consisted of a large number of very small particles
covering the surface of individual cells almost completely. Many particles seemed to adhere to the cells, however many were
floating in the medium and exhibited Brownian motion.The large number of
particles and, possibly, Brownian motion may be responsible for the image being
of poor quality. Precipitate formed within a 5 min reaction time consisted of fewer but
larger particles. None of these particles were floating. After 40 min the
precipitate consisted of even fewer particles, some of them as big as the cells themselves. The highest transfection efficiency
correlated with the generation of many very small particles.
Multiple transfections, done on the same day with the same solutions usually
give reproducible results but the transfection efficiency can vary dramatically
if experiments are performed on different days (own observations and personal communications). This provided
the motivation to systematically search for parameters which affect the
precipitate formation and would change the efficacy in transfections. Chen and co-authors (
5
) reported that the plasmid concentration needs to be optimized to achieve high
transfection efficiencies. We verified this observation and found that the
amount of DNA can have a major effect on the precipitation reaction (Fig.
3
A). At 25 [mu]g/ml all the DNA was bound to the forming precipitate within 1 min. Higher
concentrations of DNA partially inhibited the formation of precipitates,
resulting in reduced amounts of DNA being associated with an insoluble
precipitate complex. A DNA concentration of 50 [mu]g/ml DNA almost completely blocked the formation of precipitates. Even after
a 20 min incubation time, <20% of the DNA was associated with a precipitate.
Spontaneous precipitation occurs only if concentrations of calcium and phosphate
are high enough to ensure supersaturation. All the data presented above suggest
that conditions which affect the solubility of `calcium-phosphate' would directly affect the nature of the precipitate complexes.
To demonstrate this we tested five different calcium concentrations between
12.5 and 250 mM in combination with 10 different phosphate concentrations
between 0.15 and 6 mM and determined the quantity of DNA remaining in solution
after a centrifugation step (Fig.
5
A). With a calcium concentration of 250 mM, the DNA was co-precipitated at a phosphate concentration of 0.5 mM or higher. When the calcium concentration was decreased, higher phosphate concentrations were needed to co-precipitate DNA. Precipitate-complex formation and binding of DNA could be initiated with each of
the calcium concentrations used, yet at 12.5 mM calcium, the phosphate
concentration had to be >= 4 mM.
Using the phosphate concentration as a tool to change the nature of the precipitate, we designed an experiment to correlate the optical
characteristics with the transfection efficiency. Sufficient volumes of
solutions were mixed to quantify the precipitation step and also to transfect two different cell lines with the same solutions on
the same day in order to assess a possible correlation. While standard pH,
calcium and DNA concentration were chosen for these experiments (at 23oC), the incubation time was reduced to 1 min. Shown in Figure
7
A are DNA-binding and precipitate formation as measured by absorption at 320 nm,
over a range of phosphate concentrations from 0.15 to 6 mM. Corresponding transient expression levels for
TNK-tPA in CHO and HEK-293 cells in relationship to the phosphate concentrations are shown
in Figure
7
B.
A series of stable transfections was performed in CHO cells. In order to
facilitate the integration of plasmid DNA into the cellular genome, all plasmids were linearized by restriction enzyme digestion. A DHFR
vector as a selectable marker was co-transfected together with the expression vector for TNK-tPA (
16
). A single plasmid mixture and cells prepared and set up on a single day were
used. Transfection protocols were deliberately altered to reflect various
precipitation reactions. When stable clones were selected, the number of emerging clones after the selection period of 2-3 weeks was found to be different for each type of transfection used.
Most importantly, conditions that generated efficient precipitate complexes for
transient expression, also resulted in the largest number of stable clones
(data not shown,
17
).
Figure
8
shows that expression levels generated transiently (2 days after exposure of DNA to the cells) correlate to the protein levels observed in pools of stable cells selected after 2 weeks. Normalized values for transient expression were plotted against normalized values
representing stable expression of pools of recombinant cells, selected under three different stringency conditions: medium free of glycine, hypoxanthine and thymidine (GHT-), or GHT- media containing either 30 nM methotrexate (MTX) or 100 nM MTX. Stable titers were normalized within each selection group. The
stringency of selection, as expected, affected the stable transfection efficiency (numbers of clones observed): higher selective pressure resulted in
fewer clones as well as in slightly smaller colony size. Linear regression of the plotted data gave correlation coefficients of 0.83 for
the values seen in the GHT- case (99% chance of correlation), 0.73 in the 30 nM MTX case (95% chance
of correlation) and 0.62 in the 100 nM MTX case (90% chance of correlation). It
is remarkable that the correlation for the 100 nm MTX case remained good, in
spite of the fact that the individual pools of stable clones consisted of only
six clones (in the one case which gave lowest transient titers) to 35 clones
(in one of the transfections which gave better transient titers). Therefore,
early assessment of a transfection, using the supernatant just 2 days after DNA
exposure, indicates whether few or many good clones are likely to emerge weeks
later.
The principle of the calcium-phosphate-DNA precipitation method, originally developed by Graham and Van
der Eb (
1
), is simple. Nevertheless the formation of an optimal DNA-calcium phosphate co-precipitate is difficult to reproduce (
18
) because multiple parameters affect the solubility of calcium and phosphate and the
appropriate window of conditions seems to be narrow.
Two interesting issues concerning the generation of a calcium phosphate-DNA precipitate complex have been discerned. An excess of soluble DNA in
a supersaturated solution of calcium and phosphate may prolong the period
during which a precipitate forms. If high enough in concentration, DNA may even
prevent precipitate formation entirely. Therefore an error in the determination
of the concentration of DNA, e.g. due to contamination with impurities which
can influence the OD
260
reading might substantially affect the formation of the precipitate. The second issue is
temperature: the solubility of hydroxyapatite (i.e. `calcium-phosphate') is higher at lower temperatures. Kjer
et al
. (
5
) showed that this influences transfections in mosquito cells. Although calcium phosphate
is considered `insoluble' in water (
19
), no precipitate at all may be formed in standard mixtures at temperatures
close to 0oC. Most protocols recommend `room temperature' for the preparation of the
transfection mixture. Our results indicate that stricter temperature controls
are appropriate for optimal results.
In order to overcome problems with the precipitation step we suggest use of
simple tools for the early testing of solutions for transfections. A separate
phosphate stock solution can be used for adjustment purposes. Multiple precipitations can be very consistent in terms of
binding the DNA (one plasmid batch) or in terms of precipitate formation as
assessed by absorption at 320 nm. Once new solutions are characterized and
optimized, the precipitation procedure can be performed in a reproducible way
and does not need to be re-optimized for each new transfection. New DNA preparations and new compositions (
20
) of supersaturated solutions can be tested quickly. The absorption at 320 nm, assessed with an aliquot of the preparation, can be used as an approach to
study the kinetics and efficacy of precipitation.
The formation of the precipitate is a dynamic process which is terminated by
transferring the precipitation mix into the culture medium and diluting the
precipitate 10-fold. The time period allowed for generation of a precipitate is an
important parameter. Subsequent to binding most of the soluble DNA into
precipitate complexes, particles tend to grow further and transfection
efficiency will be reduced. Theoretically, the standing time can be optimized
for each set of parameters, as has been pointed out by Mahoney and Adams (
6
). In most cases a 1 min standing time appears to be a practical reaction time
and the phosphate concentration should be used as a parameter to optimize and
to control the precipitation step.
We show that concentrations of calcium, phosphate and DNA as well as temperature
and reaction time affect the formation of DNA-hydroxyapatite particles in a profound way. Most importantly,
transfection efficiency and expression levels in both transient and stable
transfections are influenced by these parameters.
The work presented here led to research evaluating the use of calcium-phosphate for DNA transfer to cells grown in stirred laboratory
bioreactors [Jordan, Köhne and Wurm (1995), submitted].
The work of M.J. was supported by the Priority Program Biotechnology (SPP),
Swiss National Foundation (SNF) (grant No. 5002-38003). The work of A.S. was supported by the Carl Duisberg Gesellschaft,
Ksln, Germany. The authors thank Dr Robert Arathoon for discussions and
encouragement and Adriana Johnson for technical assistance.
*To whom correspondence should be addressed at present address: Department of
Chemistry, Swiss Federal Institute of Technology Lausanne, 1015 Lausanne,
Switzerland
+
Present address: FH Mannheim, Hochschule für Technik und Gestaltung, 68163 Mannheim, Germany
REFERENCES
Return