SV 40-Transformed Human Corneal Keratocytes : Optimisation of Serum-free Culture Conditions

Aiming at the replacement of animal experiments in eye irritation testing, we have established a multilay ered cornea model comprising the co-culture of all three corneal cell types. It was the objective of this study to optimise serum-free culture conditions to preserve both growth and phenotype of an SV40-immortalised human corneal keratocyte cell line (HCK). Our results revealed that HCK continue to proliferate in both monolayer cultures as well as after seeding in a collagen matrix and resemble primary corneal keratocytes in morphology and functional characteristics under defined serum-free conditions. Furthermore, HCK were shown to transform into activated corneal fibroblast phenotypes in response to serum and TGF(beta)1. In summary, HCK cells mimic their in vivo (primary) precursors, both in sustaining the quiescent keratocyte phenotype (serum-starved conditions) and in responding to growth factor stimulation. Hence, this cell line may provide a useful tool to study the toxicity and wound healing response of corneal keratocytes in vitro.


Introduction
Currently, there is no accepted alternative regulatory method to eliminate the use of animals in ocular irritation testing for the risk assessment of chemicals, pharmaceuticals and cosmetics.According to an ESAC statement sufficient data support the use of the Bovine Corneal Opacity and Permeability (BCOP) test method, and the isolated chicken eye (ICe) test method as screening tests to identify substances as ocular corrosives or severe irritants in a tiered-testing strategy.Very mild irritants may soon be identified by commercially available, three-dimensional models of the corneal epithelium.However, the gap between mild and moderate eye irritation potential has to be closed by new alternative methods and the call for a mechanistically based method that considers the depth of the corneal injury and hence covers the whole severity range of toxic reactions, demands an implementation of corneal structures, such as stroma and endothelium, in an organotypic in vitro model of the human cornea (Maurer et al., 2002;Maurer et al., 2001;Jester et al., 2001;Jester, 2006).Most ocular irritants cause progressive injury to the cornea with stromal injury occurring beneath the areas of epithelial denudation.Clear exceptions to this rule are some chemicals (Maurer et al., 2001) which are more harmful to the stroma than to the epithelium.this observation highlights the need to include both epithelial and stromal components in an in vitro model as focusing only on the epithelium would underestimate the injury occurring with certain chemicals like for example hydrogen peroxide.Assays using only stratified epithelium are restricted to the detection of those materials causing only slight irritation, whereas the inclusion of stroma and stromal keratocytes broadens the range of potential irritation that can be detected.
Aiming at the establishment of a multilayered cornea model to replace animal experiments in eye irritation testing, our work focuses on the optimisation of an existing full-thickness-model of the human cornea constructed from three SV40-immortalised human cell lines (engelke et al., 2004;Zorn-Kruppa et al., 2005) with regard to standardised serum-free culture conditions for all the three different cell types of this tissue engineered cornea.they are grown in co-culture for several days, and some serum components are known to stimulate keratocyte differentiation into the contractile myfibroblast type.In this context, the SV40immortalised human keratocyte cell line (HCK) (Zorn-Kruppa Received 8th September 2008, received in revised form and accepted for publication 8th January 2009

Summary
Aiming at the replacement of animal experiments in eye irritation testing, we have established a multilayered cornea model comprising the co-culture of all three corneal cell types.It was the objective of this study to optimise serum-free culture conditions to preserve both growth and phenotype of an SV40-immortalised human corneal keratocyte cell line (HCK).Our results revealed that HCK continue to proliferate in both monolayer cultures as well as after seeding in a collagen matrix and resemble primary corneal keratocytes in morphology and functional characteristics under defined serum-free conditions.Furthermore, HCK were shown to transform into activated corneal fibroblast phenotypes in response to serum and TGFβ1.In summary, HCK cells mimic their in vivo (primary) precursors, both in sustaining the quiescent keratocyte phenotype (serum-starved conditions) and in responding to growth factor stimulation.Hence, this cell line may provide a useful tool to study the toxicity and wound healing response of corneal keratocytes in vitro.

Cell culture
the immortalised human corneal keratocyte (HCK) cell line (Zorn-Kruppa et al., 2004) was derived from primary human keratocytes transfected with a mammalian expression vector containing genes encoding for the SV40 early region.
to study cell proliferation and cell phenotype under serumfree conditions, HCK were adapted to keratinocyte growth medium (KGM ® ) supplemented with calcium chloride to a final concentration of 0.5 mM (KGM-Ca).HCK were adapted to serum-free media for at least ten and maximum fifteen passages before they were used for further tests.

Preparation of three-dimensional stromal collagen matrices
The collagen matrix for the construction of an artificial stroma was prepared according to Zorn-Kruppa (Zorn-Kruppa et al., 2005) with minor modifications.Briefly, rat tail type I collagen was dissolved in sterile 0.1% acetic acid (pH 3.3).Five volumes of the acid-soluble collagen solution were gently mixed with one volume of ten-fold concentrated F99 medium, one volume of reconstruction buffer (2.2 g of NaHCO 3 and 4.77 g HePeS in 100 ml 0.05 N NaOH) and one volume of FCS containing 1x10 5 HCK cells.tissue culture polycarbonate membrane inserts (0.4 µm pore size and 10 mm diameter; Nunc, Wiesbaden, et al., 2004) was characterised with regard to the preservation of cell specific features in spite of the SV40-immortalisation, culture medium and its serum content.
this study presents cell proliferation kinetics and phenotypical alterations of the HCK cell line (Zorn-Kruppa et al., 2004) under serum-free conditions in monolayer cell culture and after incorporation into a collagen gel matrix which reflects the stromal architecture.In addition, the response of HCK to TGFβ1 was studied together with the functional capacity of this cell line to contract the collagen matrix as a result of myofibroblast differentiation.
In the intact cornea in vivo, keratocytes are mitotically quiescent cells that establish a dendritic morphology (Fini, 1999).After isolation, these cells can maintain a dendritic, keratocyte morphology in vitro when grown in the absence of serum (Kawakita et al., 2006), while differentiation into the myofibroblast phenotype can be induced by the addition of serum, TGFβ or certain growth factors (Beales et al., 1999;Jester et al., 1996;Jester and Ho-Chang, 2003;Masur et al., 1996).Serum alone can exert a detrimental effect on the maintenance of the keratocyte phenotype due to its TGFβ content, which is known to stimulate keratocyte transformation into fibrotic cell types (Borderie et al., 1999;Masui et al., 1986).Alterations in the cellular phenotype become apparent as changes in cell shape, actin reorganisation, the expression of focal adhesion proteins and the formation of adherens or gap junctions, respectively (Fini, 1999;Fini and Stramer, 2005).the extent of transformation from keratocytes to myofibroblast-like cells is indicated by the development of prominent microfilament bundles containing α-smooth-muscle-specific actin (α-SMA) (Jester et al., 1996;Jester et al., 1999).While the quiescent phenotype is non-contractile, environmental factors that lead to phenotypical alterations can also cause functional changes regarding the cellular capacity to contract the extracellular matrix (Jester and Ho-Chang, 2003).Cellular quiescence is obtained and preserved exclusively in serum-free media, resulting in the keratocyte phenotype (Berryhill et al., 2002;Kawakita et al., 2006).
In the present study we evaluated whether the SV40-immortalised keratocytes maintain quiescent keratocyte characteristics under appropriate culture conditions and respond to serum and TGFβ stimulation like their in vivo (primary) precursors.
Serum-free collagen matrices were constructed by mixing two volumes of KGM-Ca medium containing 1x10 5 HCK, which had been adapted to serum-free medium, with one volume of reconstruction buffer and five volumes of the acid-soluble collagen solution.

Analysis of cellular proliferation
Cell proliferation was determined by Fluorescence Activated Cell Sorting (FACS).Cells were seeded in 24-well plates at a density of 5x10 3 /well, changing medium every second day.For FACS analysis cells were completely detached by incubation with a solution of 0.25% trypsin and 0.2% eDtA (200 µl/ well) at 37°C 5% CO 2 for five minutes.Cells were resuspended in 500 µl medium and transferred to an ice cooled FACS tube.Samples were vortexed carefully just prior to the measurement.the experiment was performed with a FACSCalibur tM Flow Cytometry System (Becton Dickinson, San Jose, CA).Data were analysed using the CellQuest software package (BD Biosciences, Heidelberg, Germany).
The proliferation of HCK in the artificial stroma was determined using the Mtt viability assay (Mossmann, 1983).On day 1, 3, 6, 10 and 14 the inserts were transferred to 6-wellplates containing 300µl Mtt (1 mg/ml) in the corresponding medium.After a three hour incubation period (37° C, 5% CO 2 ) the solvent was aspired, gels were rinsed three times with PBS, dried by blotting on sterile paper and transferred to a 24-wellplate.Formazan was extracted by 2 ml isopropanol for 24 hours at room temperature in the dark.the plates were sealed with parafilm to avoid solvent evaporation.Then the inserts were pierced with an injection needle to let the extraction solution run into the corresponding well, and the plate was shaken until the solution was of homogeneous colour.the optical density was measured at 570 nm.Absorbance increased linearly with cell number as quantified by counting under the microscope.

Immunofluorescent microscopy
HCK cells were plated at 5 x 10 4 /chamber on four-well labtek tM II Chamberslides (Nunc, Wiesbaden, Germany) and grown in either serum-containing or serum-free medium up to 80% confluency.Cells were exposed to 2 ng/ml TGFβ1 for three days in order to stimulate them to differentiate into myofibroblasts.For immunostaining the cells were washed twice in cold PBS containing 0.5 mM CaCl 2 and 1 mM MgCl 2 , fixed in 4% paraformaldehyde (in PBS) for 10 minutes and permeabilised with 0.1% triton x-100 for 5 minutes at room temperature.Cells were treated with Normal Goat Serum Blocking Solution IHC for 30 minutes and incubated with anti α-SMA mouse monoclonal antibody to identify myofibroblast differentiation.Indocarbocyanin-labeled anti-mouse IgG (Cy3) was used to visualise the primary antibodies.F-actin-containing stress fibers were visualised with Oregon Green ® 488-labeled phalloidin.each step was separated by 3 x 5-minute washes with PBS.
The cells on the coverslips were fixed by a Mowiol tM mounting medium and viewed with an Olympus BX41 fluorescence microscope (Olympus, Hamburg, Germany).Pictures were taken with a Nikon DxM1200 digital camera (Nikon, NY, USA).Appropriate emission filter settings and controls were included to reduce background fluorescence.
For Western blot analysis of α-SMA expression, lysates of the stimulated and non-stimulated HCK were separated by SDS/ PAGe.Gels were blotted overnight onto PVDF-membranes.After blocking with 5% non-fat dry milk in tBS-tween 20 (0.1%) for one hour at 4°C, membranes were incubated with monoclonal α-SMA antibody at a 1:12,500 dilution for 40 minutes at room temperature.the blots were washed three times in tBS-tween 20 followed by incubation with secondary antimouse antibody conjugated to horseradish peroxidase (HRP) for 1 h at room temperature.Immuno-complexes were detected using an enhanced chemiluminescence detection method.each experiment was performed three times in duplicate.
For the detection and semi-quantification of the α-SMA expression of HCK in the collagen matrix, cells were released from the gel by collagenase digestion after a culture period of six days.For this purpose, the gels were detached from their inserts and incubated in a 10:1 mixture of collagenase (2.5 mg/ml) and 0.5% trypsin at 37° C for 2 h on a shaker.After two rounds of centrifugation and washing in cold PBS, cells were subjected to lysis, protein was extracted from the cell pellet and Western blot analysis was performed as described above.

Collagen matrix contraction
Serum-containing and serum-free stromal equivalents were prepared as described above by incorporating the appropriate HCK density into the collagen matrix.After three days of submerged culture, part of the tissues were stimulated with TGFβ1 (2 ng/ ml) for another three days.Collagen gel contraction was determined on day six.Photographs of the gels were taken using an Olympus SZx12 stereomicroscope with a ColorView IIIu digital camera and cell^A image analysis software (Olympus, Hamburg, Germany).the area of each gel was determined with the AxioVision software programme from Zeiss (Carl-Zeiss, Jena).cent, non activated phenotype and the typical cortical F-actin organisation, lacking filament assembly and α-SMA staining (Fig. 2A).In contrast, both TGFβ stimulation and serum-containing culture conditions caused transformation into fibroblast-like cells together with loss of the dendritic keratocyte shape and development of a very large, spread morphology.those features are an indication for myofibroblast differentiation, which was verified by a positive staining of α-SMA (Fig. 2B, C).
The immunocytochemical results were confirmed by blotting proteins against antibodies specific for α-SMA (Fig. 3).Almost no myofibroblast-specific α-SMA is expressed in the absence of serum (lane 3).In contrast, both the addition of serum and TGFβ1 induced α-SMA expression (lane 1, 2, 4).
We also investigated the time dependence of the TGFβ1 induced phenotypic alteration in HCK cells.For this purpose, we followed the expression of α-SMA after addition of TGFβ1 on the day of seeding by Western blot analysis over seven days.As expected, HCK cultivated in serum-containing medium appeared differentiated to a certain extent, even without TGFβ1 stimulation.Addition of TGFβ1 resulted in a further increase in α-SMA expression with a maximum at day three and four (Fig. 4A).As already shown by the immunocytochemistry results, serum-free cultured HCK exhibited almost no inherent myofibroblast differentiation, but were responsive to TGFβ1 stimulation with maximum α-SMA expression on day four after seeding and stimulation, respectively (Fig. 4B).

Phenotype of HCK embedded into a 3D collagen gel matrix
The phenotype of HCK embedded in the artificial stromal matrices was also assessed by means of α-SMA expression (Fig. 5).HCK cultured under serum-free conditions for six days reflected an undifferentiated phenotype without α-SMA expression (lane 1).Lane 2 shows the profound augmentation of α-SMA in these cells reflecting a strong myofibroblast differentiation as a response to TGFβ1 stimulation of the collagen culture.Under serum-containing conditions, HCK appeared to be partially dif-the average contraction was calculated from the decrease of the gel's surface in relation to the gel surface after plating.

HCK proliferation in monolayer and 3D collagen matrix
the proliferation kinetics of HCK have been studied in several serum-free media.Most of the media supported the proliferation of cultured HCK cells, but calcium-supplemented KGM was the most efficient medium for prolonged cultivation of the cell line without serum.Figure 1A illustrates that proliferation kinetics of HCK plated in serum-supplemented F99 routine medium or in serum-free KGM-Ca medium are comparable.
HCK cells showed a very different proliferation profile in 3D collagen cultures compared to monolayers (Fig. 1A and 1B). Figure 1B shows cell growth behaviour of HCK embedded in either serum-supplemented or serum-free collagen matrices for 14 days.Apparently, HCK under serum-free conditions stop proliferation after 6 days in 3D culture.A contact inhibition effect may support the differences between the different conditions (matrix vs. monolayer).However, HCK adapted to serumstarved conditions proliferated to a much higher extent than serum-supplemented HCK in 3D cultures.this inhibition effect observed in serum-supplemented collagen matrixes may result from gel contraction, which reduces inter-cellular distances.

HCK phenotype in monolayer culture
the alterations in the phenotype of keratocytes cultivated in different media were changes in cell shape, actin reorganisation and α-smooth muscle actin (α-SMA) expression.Figure 2 shows HCK cultivated in either serum-free KGM-Ca (Fig. 2A), after stimulation with TGFβ (Fig. 2B) or in routine serum-containing F99 medium (Fig. 2C).Adaptation of HCK to serum-free KGM-Ca medium resulted in the reduction of cellular differentiation, as shown by the predominantly stellate morphology of a quies- age contraction under different conditions was calculated from the decrease of the gels' surface in relation to the average gel surface after plating.Collagen matrix contraction confirms the α-SMA expression in the stromal cells.

Discussion
the objective of this study was to formulate a medium for monolayer culture and 3D collagen gel matrix population to optimise both growth and restoration of the keratocyte phenotype of the cell line HCK, which was derived from human keratocytes by SV40-transfection. the immortalisation of ferentiated without growth factor (lane 3), hence, the increase in α-SMA expression after addition of TGFβ1 was less obvious (lane 4).

Collagen matrix contraction as a result of myofibroblast differentiation
We studied whether the phenotypical changes in HCK were paralleled by an altered contractile behaviour.Photographs of serum-free and serum-containing collagen matrices after a six day incubation period show that the presence of serum and/ or TGFβ1 induced a remarkable gel contraction (Figs.6B, C,  D).In contrast, gel contraction was not observed in tissue cultures of HCK under serum-free conditions (Fig. 6A). the aver-  Three days stimulation with 2 ng/ml TGFβ1 resulted in dramatic gel contraction in the serum-deprived (B) and serum-containing collagen matrices (D).The average collagen matrix contraction (mean ± SD) under different conditions was calculated from the decrease of the gel's surface area as % of the gel surface after plating.Each experiment was conducted with 3-6 tissues.The average contraction and standard deviation (SD) was calculated using weighted arithmetic means.In summary, SV40-transformed HCK continued to proliferate in serum-free medium while maintaining the non-contractile state.They were able to respond to TGFβ1 stimulation by differentiating into a contractile myofibroblast phenotype.This completely differentiated phenotype was reversible by the adaption to serum-free culture conditions.Despite the ability of HCK to take on keratocyte-like characteristics, the question remains whether this state represents quiescent keratocytes or repair fibroblasts.Fini (Fini, 1999) has emphasised that once the keratocyte is activated to the repair fibroblast phenotype it can further differentiate to the myofibroblast.Despite this transformation, activated corneal fibroblasts and myofibroblasts are not terminally differentiated, but have the ability to transition into one another (Maltseva et al., 2001).Over time, repair fibroblasts gradually become more quiescent and take on a keratocyte-like appearance, while biochemically, they continue to resemble fibroblasts.Hence, definitive classification of HCK cells awaits further analysis and exploration of whether these cells might be capable of re-differentiating into quiescent keratocytes.
In conclusion, our results reveal that the HCK cell line mimics in vivo keratocytes in several aspects: HCK resemble the quiescent keratocyte phenotype under serum-starved conditions and shift to activated phenotypes in response to serum or TGFβ1.Hence, this cell line may provide a useful tool to study the responses of corneal fibroblast cells to diverse environmental factors in vitro.Furthermore, the serum-free stromal construct presents an ideal component of a mechanistic in vitro wound-healing model.corneal fibroblasts using SV40-infection of rabbit keratocytes has already been reported (Barry-lane et al., 1997).the problem therein was that this transformed cell-line exhibited a myofibroblast phenotype with constitutive expression of α-SMA that could not be upregulated upon addition of TGFβ.General problems with the use of the SV40 oncogene are a possible chromosomal instability after immortalisation, block of downstream cell differentiation (Hoffschir et al., 1992) as well as loss of contact inhibition and growth control (Wilson et al., 1993;1995).
Our results show that the HCK cell line can be induced to grow and differentiate in culture to both keratocyte-like and myofibroblast-like phenotypes.These cell types can be distinguished based on cell shape, structural organisation and contractile behaviour (Fini, 1999;Jester and Ho-Chang, 2003;Jester et al., 2003).Under serum-free conditions, HCK continued to proliferate and could be subcultured for at least 10 passages while maintaining a non-contractile, keratocyte phenotype.the presence of serum and/or TGFβ induced dramatic alterations in morphology, biophysical features and function.these factors are known to promote myofibroblast differentiation, which finally results in collagen gel contraction (Borderie et al., 1999;Masui et al., 1986).
When grown in the absence of serum, HCK appeared similar to quiescent corneal keratocytes in vivo, including the maintenance of a stellate/ dendritic morphologic appearance with poorly organised microfilament bundles.The inability to contract extracellular matrix under serum-free conditions was another feature of HCK cells, which went along with a lack of α-SMA expression.This finding was consistent with previous reports on primary stromal cells (Kawakita et al., 2006), demonstrating the preservation of a keratocyte-like phenotype.Addition of TGFβ1 to serum-free cultures enhanced α-SMA expression in HCK along with the development of the typical large-spread morphology of in vitro myofibroblasts (Fini, 1999;Jester et al., 2003;Jester et al., 1995;Masur et al., 1996).this phenotypic shift is consistent with the previously reported TGFβ-induced changes in morphology and the upregulation of α-SMA in cultured primary keratocytes (Assouline et al., 1992;Jester et al., 1996;Jester et al., 1999;Petridou et al., 2000).Moreover, in the presence of TGFβ1, HCK became capable of contracting collagen gels, demonstrating the conversion of keratocytes to myofibroblasts along with the loss of keratocyte-like characteristics (Assouline et al., 1992;Jester and Ho-Chang, 2003;Kurosaka et al., 1998).
In serum-containing standard medium HCK expressed myofibroblast-like features even without TGFβ1 stimulation.The addition of 5% FCS caused a loss of the dendritic morphology in part of the HCK.In accordance, α-SMA assembly was clearly detectable along intracellular stress fibers marking cellular transformation into a myofibroblastic phenotype (Funderburgh et al., 2001).In serum-containing three-dimensional collagen gels HCK were capable of matrix contraction, even without TGFβ1 stimulation.In summary, 5% FCS exerted a universal detrimental effect on the maintenance of the keratocyte phenotype.Hence, building an organotypic corneal construct from serum-fed HCK is not possible.

Fig. 4 :
Fig. 4: Western blot analysis of time dependent α-SMA expression.TGFβ1 was added on the day of seeding (day 0).In both serumfree and serum-containing medium, HCK are sensitive to TGFβ1 stimulation.Under serum-containing conditions cells showed anti-α-SMA staining already on day one (A).α-SMA expression increased with time until day three.Under serum-free culture conditions (B), HCK showed maximum protein expression on day four.

Fig. 6 :
Fig. 6: Collagen matrix contraction assay showing the contractile properties of HCK in serum-free (A and B) and serumcontaining collagen gels (C and D).Three days stimulation with 2 ng/ml TGFβ1 resulted in dramatic gel contraction in the serum-deprived (B) and serum-containing collagen matrices (D).The average collagen matrix contraction (mean ± SD) under different conditions was calculated from the decrease of the gel's surface area as % of the gel surface after plating.Each experiment was conducted with 3-6 tissues.The average contraction and standard deviation (SD) was calculated using weighted arithmetic means.

Fig. 2 :
Fig. 2: Stellate morphology of the quiescent, non activated phenotype of HCK cells with typical F-Actin organisation in serum-starved medium (A).Myofibroblast differentiation induced after TGFβ stimulation (B) or addition of serum (C) as shown by anti-α-SMA staining.