Sumoylation Dynamics During Keratinocyte Differentiation

HaCaTs, a naturally immortalized human keratinocyte cell line, express the SUMO machinery

The SUMO system has been extensively studied, and the enzymes required for sumoylation are widely expressed in many tissues (Chen et al., 1998; Johnson, 2004). However, the sumoylation system remains largely unexplored in human keratinocytes, an important model for differentiation. In order to study the functional importance of this modifier in the context of keratinocyte differentiation, we first developed culture conditions to maintain HaCaT cells stably in either a basal or differentiated phenotype. HaCaT cells cultured in 0.03 mM calcium were spindle-shaped and loosely packed (Figure 1A). In addition, there was little or no expression of two differentiation markers, keratin 1 (K1) and involucrin (Inv), at the RNA (Figure 1B) or protein (Figure 1C) level, consistent with these cells remaining in a basal like state. In contrast, HaCaT cells maintained in 2.8 mM calcium became more cuboidal, developed cell-to-cell tight junctions, and expressed both differentiation markers, K1 and Inv, at the RNA and protein level (Figures 1A–C). Under both high and low calcium conditions, the cultures could be maintained for many passages without any subsequent changes in morphology or marker expression pattern. Therefore, these growth conditions allow HaCaT cells to be stably maintained in vitro in two alternative states resembling pre- and post- differentiated skin cells.

To characterize the sumoylation system in human keratinocytes, basal and stably differentiated HaCaT cultures were evaluated for expression of sumoylation components at both the RNA and protein levels. RT-PCR and quantitative RT-PCR (Q-PCR) analyzes were used to detect transcripts for all the core SUMO components, including SUMO1, SUMO2, SUMO3, SAE1, SAE2, and Ubc9 (Figures 2B and 2C), as well as the differentiation marker involucrin (Figure 2A). Additionally, one SUMO ligase (Miz-1 or PIASxβ) and one SUMO protease (SENP1) were examined by RT-PCR. Consistent with the qualitative results shown in Figure 1B, Q-PCR detected approximately 14-fold higher levels of involucrin mRNA in the high calcium-maintained keratinocytes compared to the low calcium culture, confirming the differentiation state of these cells (Figure 2A). By RT-PCR, transcripts for all of the SUMO component genes were detected in both the basal and differentiated HaCaT cells, indicating that the relevant genes for an active sumoylation system were actively transcribed in both cell populations (Figure 2B). Although the HaCaT cells were maintained in two distinct phenotypic states, the transcript levels for the sumoylation components tested were similar under both culture conditions (Figure 2C). Immunoblot analyses confirmed the expression of Ubc9 and SAE1 proteins, and the levels of each of these proteins were only slightly higher in the differentiated cells compared to the undifferentiated HaCaT cells (Figure 2D). Next, we tested if the sumoylation system was active in HaCaT cells as evidenced by the presence of sumoylated proteins. Immunoblotting of total cell extracts using a polyclonal anti-SUMO antibody showed that sumoylation occurred in both the basal and differentiated cell populations (Figure 2E). Although the differences in expression levels of the sumoylation components were small, the sumoylation patterns in differentiated versus basal cells were slightly different: bands at ~40, 60, and 95 kDa were intensified in high calcium cells compared to basal cells (Figure 2E, arrows). Since the basal and differentiated cell cultures were both equally proliferative and were generated from a single stock of HaCaT cells, we believe that these changes in sumoylation pattern reflect differentiation-related events rather than intrinsic differences in genetic background or growth capacity of the two populations of cells.

SUMO and Ubc9 are expressed in stratified human foreskin keratinocytes

To corroborate the HaCaT observations, the expression of SUMO and Ubc9 was evaluated in primary human foreskin keratinocytes (HFKs) that were stratified in organotypic cultures. Immunohistochemistry analyses were performed with an anti-SUMO polyclonal antiserum and a purified polyclonal antibody against Ubc9, and antibody specificity was confirmed by blocking experiments using purified SUMO1 or Ubc9 (Figure 3). SUMO was detected in the nucleus and cytoplasm of keratinocytes found in all layers of the epithelium (Figure 3C). However, as this antiserum cross reacts with SUMO2/3 (unpublished observations), which SUMO types were present and whether or not there were changes in type expression in different layers is unclear. In contrast, Ubc9 was barely detected in the basal layer of the epithelium where the cells remained relatively unstained (Figure 3A, arrowheads). Strong anti-Ubc9 staining was detected in the intermediate differentiated layers just above the basal layer, and Ubc9 was present in both the nucleus and cytoplasm of the keratinocytes in this region (Figure 3A, black arrows). Although Ubc9 could be detected in nuclei throughout the upper layers (Figure 3A and data not shown), its overall expression faded as keratinocytes moved closer to the outer layer of the epithelium. It appears from these results that Ubc9 expression is low in basal cells, transiently increases with the initiation of differentiation in the suprabasal layers, and declines as terminal differentiation progresses. As Ubc9 is the only E2 enzyme for sumoylation, we speculate that the pool of sumoylated protein might be dramatically altered as well during this process. Interestingly, Ubc9 expression is maximal in the layers just below where keratin 1 expression commences (compare Figures 3A and 3E). Overall, these analyses demonstrate that sumoylation components are present in stratified human skin epithelium and that Ubc9 levels change during differentiation. The similar observations in HaCaT cultures suggest that monolayer HaCaT cells are a good representative model of the sumoylation system in normal human keratinocytes.

Calcium induced differentiation in HaCaT cells up regulates the sumoylation system

Although only minor differences in expression of the sumoylation components were detected between the two steady-state populations of HaCaT cells, the immunohistochemistry results suggested that more substantial changes in sumoylation might be taking place during the process of differentiation. To investigate this possibility we examined the sumoylation system in HaCaT cells undergoing active calcium-induced differentiation. Addition of calcium to the basal HaCaT cultures triggered a rapid response resulting in repression of the basal keratin markers, morphological changes (data not shown), and mRNA expression of differentiation markers such as K1 and Inv within 3 to 4 days (Figures 4A and 4B). The Inv transcript detected by Q-PCR progressively increased from 48 to 144 hours following calcium induction (Figure 4A). Likewise, the K1 transcript, which was completely repressed in undifferentiated (0 hours) HaCaT cells, was detected by RT-PCR at 72 hours post calcium induction, reaching a maximum expression by 120 to 144 hours (Figure 4B). No K1 expression was detected in uninduced cells for up to 120 hours (Figure 4B). Note, however, that the faint expression of K1 in uninduced cells by 144 hours was the result of the HaCaT cells reaching high confluency at this late time point (Capone et al., 2000). Based on K1 and involucrin expression, complete in vitro differentiation of HaCaT was reached within 144 hours after calcium-induction. Therefore, we chose to limit our study to a six-day period during which we monitored the regulation of the sumoylation system while the HaCaT cells progressively differentiated. RT-PCR and Q-PCR showed rapid changes in the transcript levels for the genes encoding many of the sumoylation components (Figures 4C and 4D). Transcripts for SUMO2, SUMO3, Ubc9, and SAE1 were up regulated 2-fold or more by 24 hours post-induction, with the SUMO2 and Ubc9 transcripts increasing by 5-fold or more at 144 hours post-induction (Figure 4C). The final increase for SUMO3 and SAE1 was more modest, reaching a maximum increase of 2–3 fold. The results for SAE2 were less clear, and the transcript levels did not consistently cross the 2-fold threshold until day 6. SENP1 transcripts detected by RT-PCR also gradually increased up to 144 hours, while Miz-1 (PIASxβ) transcripts were expressed maximally at 48 hours post induction and then declined (Figure 4D). We conclude from these observations that there is a general upregulation of transcription of the genes for the sumoylation components accompanying the early stages of differentiation. Interestingly, the SUMO1 gene responded differently to calcium induction than did the SUMO2 and 3 genes (Figure 4E). SUMO1 transcript levels did not increase significantly during the first 5 days post-induction with calcium, and even at day 6 the transcript levels were barely at the 2-fold threshold. Thus, SUMO1 gene expression during calcium-induced differentiation appears to be regulated differently than the genes for the other SUMO components, though the biological significance of this result is not yet defined.

Transcriptional upregulation of the sumoylation system upon calcium signaling is accompanied by an increase in the corresponding proteins and an increase of overall sumoylation in HaCaT cells

Increase of specific transcripts during a biological process is not necessarily functionally important unless it is followed by a corresponding increase in protein expression. Immunoblot analyses of representative sumoylation components indicated that the Ubc9, SAE1, and SENP1 proteins levels remained relatively constant in uninduced cultures over the 6-day incubation period (Figure 5, B and C). In contrast, all three of these proteins were rapidly, but transiently, upregulated by more than 3-fold during HaCaT cell differentiation (Figure 5, B and C). Maximum protein expression was reached at 48 to 96 hours post induction and then declined. Although the absolute increases for SAE1, Ubc9 and SENP1 were slightly different, the overall expression pattern was similar for all 3 enzymes (Figure 5C). Interestingly, while protein expression levels declined after 96 hours post induction, transcripts were still elevated and had not begun to decline during the time period examined (Figure 4C). The explanation for this discordance is not known, but may reflect the existence of a posttranscriptional regulatory mechanism that controls the level of protein expression for the different components of the sumoylation system during keratinocyte differentiation. Nonetheless, the combined results in figures 2, 4 and 5 indicate that calcium triggers a rapid and coordinated activation of transcription for many SUMO system genes leading to a corresponding accumulation of the SUMO system proteins, eventually followed by a decline to near basal protein levels. The observation in HaCaT cells that there is a transient upregulation of Ubc9 expression strongly correlates with the results from the Ubc9 immunohistochemistry data in the stratified HFKs (Figure 3) which showed a transient increase in Ubc9 protein levels in the suprabasal layers that declined with further differentiation. Importantly, as in the stratified HFKs, the peak expression of Ubc9 in differentiating HaCaT cells occurred just prior to expression of the K1 protein (Figure 5A–C).

Next, we wanted to test if the accumulation of sumoylation components increased the overall population of sumoylated targets. Total cell extracts of calcium-induced HaCaT cells were collected every 24 hours up to 144 hours post-induction and were analyzed by immunoblotting (Figure 6). Overall sumoylation was clearly increased between 96 and 144 hours and discernible changes in the sumoylation of specific substrates were observed. During the differentiation process, several substrates (indicated by arrows) showed increased sumoylation, while a major sumoylated product at about 70 kDa decreased (arrowhead). In addition, new sumoylated bands appeared at about 45 and 95 kDa (arrows). No changes in substrate sumoylation were observed in parallel cultures maintained in low calcium medium (data not shown). Enhanced sumoylation during HaCaT differentiation was even more pronounced when the samples were evaluated by 2D gel electrophoresis (unpublished observations). These results indicate that the increased expression of the sumoylation components leads to increased sumoylation activity, and that calcium-induced differentiation of keratinocytes is accompanied by dynamic changes in the pattern of sumoylated proteins.

Inhibition of the sumoylation system perturbs the HaCaT cell differentiation process

To investigate the functional importance of sumoylation in the process of keratinocyte differentiation, we infected basal HaCaT cells with an adenovirus expressing c-Myc tagged Gam1 (Ad-Gam1) and then exposed the cells to high calcium medium to trigger differentiation. Gam1 is known to inhibit sumoylation by inducing degradation of SAE1/2 and Ubc9 (Boggio et al., 2004). By 24 hours post-infection, Gam1 was detected in HaCaT cells infected with Ad-Gam1, but not in cells infected with Ad-GFP (Figure 7B). In the Ad-GFP infected cells, the Ubc9 initially present increased 3.2-fold by 48 hours post calcium induction, consistent with the response seen in the absence of viral infection (see Fig. 5B). In contrast, cells expressing Gam1 showed diminished Ubc9 levels by 24 hours and no detectable Ubc9 by 48 hours when Gam1 levels had reached maximal expression. A similar effect was seen in co-infections with Ad-Gam1 and Ad-GFP, indicating that Ad-GFP had no stabilizing effect on Ubc9. In contrast to these infection studies where the transgene could be delivered effectively into nearly 100% of the cells, alternative attempts to inhibit sumoylation via transfection approaches were unsuccessful due to the poor transfection efficiency of HaCaT cells in low calcium medium (unpublished results).

When examined microscopically, HaCaT cells expressing Gam1 exhibited a disruption of the normal morphological changes associated with differentiation. By 72 hours after Ad-Gam1/Ad-GFP co-infection and calcium induction, when Ubc9 was absent, the differentiating HaCaT cells showed dramatic cellular rearrangements, including lose of tight cell-cell junctions and typical cuboidal shape (Figure 7A). This abnormal morphology was not observed in cells infected with Ad-GFP alone at any time during the differentiation process (Figure 7A). We also observed in the co-infected cultures that GFP expression increased throughout the experiment, while in cells infected with Ad-GFP alone the GFP expression gradually diminished and was almost undetectable by 120 hours post induction. This observation was consistent with Gam1’s known enhancing effect on transcription (Chiocca et al., 2002), and further demonstrated Gam1’s activity in HaCaT cells. Identical morphological abnormalities were also observed in cultures infected with Ad-Gam1 alone (data not shown). Additionally, when basal cultures were infected with Ad-Gam1 without subsequent calcium induction there was no change in cell morphology, while stably differentiated HaCaT cultures were susceptible to Gam1 (Supplemental Figure 1). The insensitivity of the basal cultures to Gam1 confirms that inhibition of sumoylation is not directly affecting growth or morphology of the keratinocytes in the absence of differentiation. In contrast, the susceptibility of the stably differentiated cells to Gam1 further links differentiation state with a sumoylation requirement.

In addition to its inhibition of sumoylation, Gam1 is also known to inhibit HDAC activity (Chiocca et al., 2002). To exclude the possibility that the observed morphological effects on HaCaT cells could be due to HDAC inhibition, we conducted the calcium induction in the presence of trichostatin A (TSA), a chemical inhibitor of HDACs (Yoshida and Horinouchi, 1999). Basal HaCaT cells were infected with Ad-GFP, induced with calcium, and then treated with 100 mM TSA at 24 hours after induction/infection. GFP expression was prolonged in the TSA-treated cultures compared to the untreated cells, consistent with TSA mediated enhancement of transcription through HDAC inhibition (data not shown). However, no abnormal morphology was observed during differentiation in TSA, indicating that HDAC inhibition does not contribute to the Gam1 effect on differentiating keratinocytes (Figure 7A).

In addition to keratinocyte morphology, the effect of Gam1 expression on the K1 differentiation marker was also evaluated. Gam1 caused a delay in and reduction of K1 expression in calcium induced cells (Figure 7C). While we cannot exclude that an unknown, off-target effect of Gam1 may also contribute to both the defects in morphology and marker expression, the simplest explanation is that these defects are related to the dramatic inhibition of sumoylation. Altogether, the results presented here suggest that modification with SUMO may be a key event for the proper timing of the steps leading to keratinocyte differentiation, and that sumoylation may be an important regulator for this biological process.

Cell Culture

HaCaT cells (kindly provided by Dr. Bokoch, The Scripps Research Institute, La Jolla, CA) were cultured in calcium-free DMEM (HyClone), with calcium-depleted 10% FBS (Gemini Bioproducts), 4 mM L-Glutamine (HyClone), and supplemented with calcium chloride to a final concentration of 0.03 mM (low calcium medium) or 2.8 mM final concentration (high calcium medium). FBS was calcium-depleted by incubation with Chelex 100 resin (Bio-Rad Laboratories, Inc.) for 1 hour at 4°C according to the manufacturer’s protocol. The Chelex was subsequently removed using a 50 ml Millipore 0.22 µm filter unit system (Millipore Corp). To obtain cells exhibiting a basal keratinocyte phenotype, HaCaT cells were cultured in low calcium medium for at least 3 weeks and were maintained in the same culture conditions thereafter. Low calcium HaCaT cultures were never allowed to exceed 85% confluency. For stably differentiated HaCaT cells, the cultures were maintained in high calcium medium for at least 3 weeks prior to analysis. HaCaT cultures in either medium remained fully proliferative with similar doubling times (unpublished observations). Human embryonic kidney 293A (HEK 293A) cells (Invitrogen Corp.) were maintained in DMEM supplemented with 10% FBS (Gemini Bioproducts). For the trichostatin A (TSA) experiment, HaCaT cells were treated with 100 mM TSA (Tocris Bioscience) 24 hours after infection with an adenoviral vector expressing GFP (Ad-GFP) and calcium induction. HaCaT cells were observed up to 144 hours post calcium induction.

RT-PCR and quantitative RT-PCR

All RNAs were extracted using the RNAqueous kit (Ambion). RNA concentration was measured using a spectrophotometer and 5 ng/µl aliquots were stored at −80°C until use. All primers were designed to overlap exon-exon junctions, therefore only amplifying the cDNAs targets and not the genomic sequences (primers sequences are provided in the supplemental Tables 1 and 2). The one-step RT-PCR mixture contained 50 ng of RNA in a final reaction volume of 25 µl. The mixture also included 2.5 µl of 10X Taq DNA Polymerase buffer (Promega), 50 units of MMLV reverse transcriptase, 1 unit of Platinum Taq (Invitrogen Corp.), 4 mM dNTPs, 20 units of RNAse OUT (Invitrogen Corp.), 6 mM of MgCl₂, and 0.4 µM of each primer. The one step RT-PCR was performed for 20 minutes at 42°C for the reverse transcription step, followed by 90 seconds at 94°C, and 40 cycles of amplification (94°C for 30 seconds; 60°C for 30 seconds; 72°C for 60 seconds). Amplifications were performed in a PTC-200 Peltier Thermal Cycler machine (MJ Research, Inc.). Amplified products were analyzed on 2.5% agarose gels and visualized with an Innotech Alphaimager system (Alpha Innotech Corp.). For the quantitative RT- PCR (Q-PCR), one step RT reactions were performed in a single well using 50 ng of harvested RNA in a 50 µl final volume. In addition to the RT components described above, reactions for Q-PCR contained 0.2 µM of LUX primers (designed with the Invitrogen custom primer software), 20 units of RNAse OUT, 2 µl of Super Script III enzyme solution (which includes MMLV-RT, Platinum Taq, and dNTPs), and 1 µl of Rox dye , all from Invitrogen Corp. The LUX β-actin primer set (Invitrogen) was used to detect the internal control gene. The PCR reaction conditions for Q-PCR were the same as the RT-PCR described above. The Q-PCR plates were read with an ABI 7500 real time PCR instrument (Applied Biosystems), and detection of the FAM or JOE label was recorded during the 72°C step. Results were graphed as the fold increase of the relative quantitative RQ values where RQ =2^(−ΔΔCt). ΔCt was calculated as the average Ct for β-actin minus the average Ct for the gene of interest, with each sample being run in duplicate. Using the ΔCt values from each time point, the ΔΔCt for each mRNA examined was calculated as the ΔCt of the time 0 sample minus the ΔCt of time X (in hours) after induction. The data shown are the average from at least 3 independent RNA preparations collected in separate experiments.

Affinity purified polyclonal antibodies against Ubc9

Rabbit polyclonal serum #12741 was produced in-house using affinity purified Ubc9 as the immunogen. A four immunization regime was followed, and 2 weeks after the final boost the rabbit was exsanguinated. To affinity purify anti-Ubc9 antibodies, 750 µg of affinity-purified GST-Ubc9 were diluted in a final volume of 500 µl using 1× PBS, and the resulting dilution was dispensed on an 82 mm diameter, 0.45 µm pore size, Protran Nitrocellulose filter (Schleicher & Schuell Inc.). The membrane was dried for 30 minutes at room temperature, then re-wetted and blocked by incubation at room temperature for 30 minutes in 15 ml of 1× PBS supplemented with 1% BSA. The blocked membrane was incubated for 2 hours at room temperature with 10 ml of a solution containing 2.5 ml of rabbit polyclonal serum #12741, and 7.5 ml of 1× PBS supplemented with 10 mg/ml BSA and 0.05% Tween 20. The membrane was subsequently washed four times each with 15 ml of 1× PBS supplemented with 0.05% Tween 20, and four additional times each with 15 ml of 1× PBS alone. The bound antibodies were eluted by incubation with 2 ml of elution buffer (100 mM glycine, pH 2.5, 0.02% NaN₃) for 5 minutes at room temperature. The eluted antibodies were neutralized with 200 µl of 1.0 M Tris-Cl (pH 8.0), aliquoted, and stored at −70°C. The purity of the affinity-purified antibodies was determined by gel electrophoresis and immunoblotting, and its reactivity against Ubc9 was determined by immunoblotting using purified Ubc9 and unfractionated mammalian cell extracts.

Immunohistochemistry

Human foreskin keratinocytes were grown in organotypic raft cultures to form stratified epithelium as previously described (Lambert et al., 2005). All chemicals used for immunohistochemistry were from Biocare Medical (Concord, CA). Slides were deparafinized in xylene and re-hydrated according to the manufacturer’s protocols. The following antibodies and dilutions were used: anti-SUMO serum #12783 (Rosas-Acosta et al., 2005a), 1:750; affinity purified polyclonal antibodies against Ubc9, 1:20; and polyclonal anti-human keratin 1 (anti-HK1, Covance, Berkeley, CA), 1:5000. Specificity control blocking experiments were performed by adding the corresponding purified proteins (25 µg of Ubc9 or 7 µg of SUMO1) to the cognate antibody dilution. The SUMO and Ubc9 proteins used in the experiments were purified as previously described (Rosas-Acosta et al., 2005a). After immunostaining, the slides were counterstained with hematoxylin.

Immunoblots and densitometry

Total cell extracts were prepared by adding a 1:1 (V/V) mixture of RIPA buffer (50 mM Tris-Cl, pH8, 150 mM NaCl, 5 mM EDTA, 15 mM MgCl₂, 1% NP40, 0.1% SDS, 1 mM DTT, 1:200 protease inhibitor cocktail and 10 mM N-ethylmaleimide) and 4 × sample buffer (100 mM Tris-Cl, pH 6.8, 20% glycerol, 8% SDS, 0.02% bromophenol blue, 4% β-mercaptoethanol) directly to the cells. The cells were shaken gently for 5–10 seconds, and the resulting lysate was collected by pipeting. Samples were heated at 95°C for 5 minutes and sonicated for 30 seconds using a Misonix sonicator 3000 (Misonix, Inc.). Samples were electrophoresed on 10% or 12.5% polyacrylamide gels and then transferred onto 0.45 µm Immobilon-P membranes (Millipore). The membranes were blocked for at least 15 minutes with 3% non-fat milk in TTBS (150 mM NaCl, 50 mM Tris-Cl, pH 7.4, 0.005% Tween 20), and incubated for 1 hour or over night with the primary antibodies listed below at the indicated dilution: rabbit anti α-tubulin (Santa Cruz Biotechnology Inc.), 1:15,000; anti-c-Myc monoclonal antibody (Santa Cruz Biotechnology Inc.), 1:500; anti-RanGAP monoclonal antibody (Zymed/Invitrogen Corp.), 1:2,500; rabbit serum #12783 against SUMO (Rosas-Acosta et al., 2005b), 1:1,000; affinity-purified polyclonal antibodies against Ubc9, 1:500; anti-human K1 rabbit serum (Covance), 1:1000; anti-SAE1 sheep serum (Axxora), 1:2,000; anti-SENP1 rabbit serum (Imgenex Corp.), 1:2,000; and anti-involucrin rabbit serum (LabVision Corp.), 1:1,000. After reaction with the primary antibodies, the membranes were incubated with Horseradish Peroxidase-conjugated antibodies (Santa Cruz Biotechnology) at 1:10,000 for one hour. The membranes were subsequently rinsed in TTBS, treated with the Western Lightning Chemiluminescence reagent (PerkinElmer Life And Analytical Sciences, Inc.), and then visualized with X-ray film. Quantitative differences were determined by densitometry using an Innotech Alphaimager (Alpha Innotech Corp.) and were normalized to the α-tubulin signal. Quantitative results are the average of at least 3 separate experiments.

Virus production and infection

C-Myc tagged Gam1 adenoviral DNA (Ad-Gam1) was kindly provided by Dr. Matt Cotten (GPC-biotech, Munich, Germany). The Ad-Gam1 DNA was transfected into HEK 293A cells using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s recommendation. Cells were lysed by freeze-thaw and the supernatant was collected. The initial viral stock was subsequently amplified with two rounds of infection on HEK 293A cells and the final freeze-thaw supernatant collected as the high-titer stock. Adenovirus expressing GFP (Ad-GFP) was kindly provided by Dr. G. Davis (Texas A&M Health Science Center, College Station, TX) and amplification was performed as above. Titer was assessed by the limiting dilution method (Qbiogene Adenovirus Manual, version 1.4) using HEK 293A cells plated at 1 × 10⁶ cells/well on 6-well plates. For adenovirus experiments, HaCaT cells maintained in low calcium medium were plated at 5 × 10⁶ in T75 flasks 20 hours prior infection. Each culture was infected at an MOI of 300 in 3 ml of calcium-free medium supplemented with 8 µg/ml polybrene (Fisher Scientific). Three hours after infection the medium was removed, cells were trypsinized, and released cells were resuspended in 10 ml of high calcium medium to induce differentiation. The resulting cells suspension was split among the wells of a 6 well plate at a ratio such that the cells in each well would achieve approximately 80% confluency by the time of collection. Cells were maintained in high calcium medium and harvested at various times post plating by direct lysis in the wells using the 1X RIPA:4X sample buffer mixture described above. Proteins were analyzed by immumoblotting as described in the previous section. Cells were visualized by phase contrast and fluorescence microscopy at a magnification of 200X using an Olympus IX70 microscope. Images were captured digitally using a Qcolor3 camera (Olympus).