Modeling Subgenomic Hepatitis C Virus RNA Kinetics during Treatment with Alpha Interferon

HCV subgenomic replicon cells and interferons.The clone B HCV sg1b Huh7 cells were obtained from the NIH AIDS Research and Reference Reagent Program and have been previously described (2). The cells were cultured in complete Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 100 mg/ml streptomycin, 2 mM l-glutamine, and 500 μg/ml Geneticin (Invitrogen). Recombinant human IFN-α-2a (PBL Biomedical Laboratories, New Brunswick, NJ) was resuspended to a concentration of 1 U/μl in complete DMEM supplemented with 10% fetal bovine serum, aliquoted into single-use tubes, and stored at −80°C.

IFN-α HCV replicon inhibition assay.To ensure that the cells would not reach confluence and require splitting during our 9-day IFN-α treatment experiment, sg1b cells were seeded at low density (∼5 × 10⁴ cells/well) in six-well tissue-culture dishes in complete DMEM. One day later, cells were mock treated or treated with IFN-α at 100, 250, 500, and 1,000 U/ml concentrations in a total volume of 2 ml of complete DMEM without Geneticin. Fresh vehicle control or IFN-α containing medium was added to the cultures every 12 h for the entire duration of the experiment to ensure continuous IFN signaling. At the indicated times (see the figures), medium was harvested from triplicate wells for cytotoxicity analysis, and RNA was isolated from cells in 500 μl of 1× Nucleic Acid Purification Lysis Solution (Applied Biosystems, Foster City, CA) for real-time reverse transcription-quantitative PCR analysis (RT-qPCR).

Cytotoxicity assay.The release of lactate dehydrogenase (LDH) from cells into the culture medium occurs when cell membrane integrity is compromised, and thus this release can be used to assess cytotoxicity. Hence, LDH levels in the medium of IFN-α-treated cultures were determined using a CytoTox-ONE Homogeneous Membrane Integrity Assay (Promega, Madison, WI), according to the manufacturer's instructions. Detection of the fluorescent resorufin product was assessed using an excitation wavelength of 560 nm and an emission wavelength of 590 nm (FLUOstar Optima fluorometer; BMG Labtech, Durham, NC). A complete cell lysis sample was run as a positive control, and medium alone was run as a negative control.

RNA analysis.Total cellular RNA was isolated using a 1× Nucleic Acid Purification Lysis Solution (Applied Biosystems) and purified using an ABI Prism 6100 Nucleic Acid PrepStation (Applied Biosystems), as per the manufacturer's instructions. One microgram of purified RNA was used for cDNA synthesis using the TaqMan RT reagents (Applied Biosystems), followed by SYBR Green RT-qPCR using an Applied Biosystems 7300 real-time thermocycler (Applied Biosystems). Thermal cycling consisted of an initial denaturation step for 10 min at 95°C, followed by 40 cycles of denaturation (15 s at 95°C) and annealing/extension (1 min at 60°C). HCV, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), IFN-stimulated gene 15 (ISG15), and ISG56 transcript levels were determined relative to a standard curve comprised of serial dilutions of plasmid containing the relevant cDNAs. The PCR primers used to detect genotype 1b HCV were 5′-CGACACTCCACCATAGATCACT-3′ (sense) and 5′-GAGGCTGCACGACACTCATACT-3′ (antisense); for human GAPDH (NMX002046), the primers were 5′-GAAGGTGAAGGTCGGAGTC-3′ (sense) and 5′-GAAGATGGTGATGGGATTTC-3′ (antisense); for human ISG15 (NM_005101), the primers were 5′-CAGCGAACTCATCTTTGCCAGTA-3′ (sense) and 5′-CCAGCATCTTCACCGTCAGG-3′ (antisense); and for ISG56 (NM_001001887), the primers were 5′-GGGCAGACTGGCAGAAGC-3′ (sense) and 5′-TATAGCGGAAGGGATTTGAAAGC-3′ (antisense).

Indirect immunofluorescence.At 5, 6, 7, and 8 days posttreatment, one four-chamber slide containing mock-treated and 100-U/ml and 250-U/ml IFN-α-treated sg1b cell cultures was fixed with 4% paraformaldehyde (Sigma), and intracellular staining was performed as previously described (10, 28). Briefly, the mouse monoclonal anti-HCV NS5A antibody E9-10 (a gift from Charlie Rice) was used at a dilution of 1:500, followed by incubation with a 1:1,000 dilution of an Alexa 555-conjugated goat anti-mouse immunoglobulin G antibody (Molecular Probes, Carlsbad, CA) for 1 h at room temperature. Cell nuclei were stained by Hoechst dye. Chambers were removed, and stained cells were coverslipped and mounted in Prolong antifade mounting medium (Molecular Probes) and visualized on a Zeiss Axiovert LSM 510. Images were generated using LSM Image Browser software and exported to Adobe Photoshop (Adobe Systems Inc., San Jose, CA).

Statistical analysis.To compare mean sg1b RNA levels at each time point between IFN-α treatments, we used the general linear model univariate analysis method. All multiple comparisons were adjusted using a Bonferoni correction, and in all cases, a P value of ≤ 0.05 was considered significant (SPSS software, version 15, Chicago, IL). To determine whether sg1b RNA levels differ between IFN-α treatment doses over time, we used Friedman's test, the nonparametric analogue of the two-way analysis of variance (S-PLUS, version 7.0.2; Seattle, WA). HCV sg1b RNA decline slopes were measured by linear regression (SPSS, version 15; Chicago, IL). Nonlinear least-squares regression using a Levenberg-Marquardt algorithm was used to fit the solutions of mathematical models to the HCV RNA decline data.

Increasing concentrations of IFN-α are not toxic but inhibit replicon cell growth.It has been reported that the growth state of the host cell can have detectable effects on the replication rate of HCV replicons (14, 23, 27). Because we wanted to ensure that our HCV IFN-α inhibition experiments were performed under conditions that limited any nonspecific effects IFN-α might have on the host cell, we initially monitored the effect that increasing IFN-α doses (i.e., 100, 250, 500, and 1,000 U/ml) had on Huh7 clone B cell viability and growth (Fig. 1). Mimicking the 9-day IFN-α inhibition assay described in Materials and Methods, replicon cells were plated at a low density (5 × 10⁴ cells/well) in six-well tissue culture dishes and 1 day later were mock treated or treated every 12 h with IFN-α at various doses for 2, 3, 4, 6, 8, or 9 days. As shown in Fig. 1, cell growth was similar in all cultures regardless of the IFN-α dose until day 6 posttreatment. At this time, however, cells treated with 1,000 U/ml IFN-α stopped dividing and remained at a cell confluence of 60% (Fig. 1). Likewise, the expansion of cells treated with 500 U/ml IFN-α slowed between days 6 to 8 and then stopped by day 8 when the cells reached 70% confluence of (Fig. 1). In contrast, cells treated with 100 and 250 U/ml IFN-α expanded at a rate equivalent to mock-treated control cultures (Fig. 1) until at least day 8 (Fig. 1).

• Open in new tab
• Download powerpoint

FIG. 1.

HCV replicon cell growth during treatment with various concentrations of IFN-α. HCV replicon cells were incubated with various concentrations of IFN-α for 2, 3, 4, 6, 8, or 9 days. Fresh IFN-α was added to the medium every 12 h. Cells were trypsinized and counted at indicated time points posttreatment. Cell number, expressed as relative confluence of the culture, was assessed as an indicator of relative cell growth rates.

To determine if any of these IFN-α treatment doses were toxic, we measured the release of LDH into the culture medium. The results indicated that none of the IFN-α concentrations tested (i.e., 100 to 1,000 U/ml) were toxic as LDH concentrations measured from treated cultures were comparable to the background LDH concentrations measured in the control cultures (data not shown).

IFN-α reduces intracellular subgenomic HCV RNA levels in a biphasic-like manner.To analyze the effect of IFN-α on HCV sg1b RNA levels over time during this 9-day treatment, parallel replicon cell cultures were treated with the same four IFN-α concentrations tested above, i.e., 100, 250, 500, and 1,000 U/ml, and sg1b RNA was measured in triplicate samples by RT-qPCR analysis after 2, 3, 4, 6, 8, and 9 days of treatment (Fig. 2). Interestingly, when treated with 100 U/ml IFN-α, sg1b RNA levels fell in a biphasic manner that consisted of a rapid first-phase RNA decline (that lasted until about day 4), followed by an extremely slow or flat second phase (Fig. 2 and Table 1). The rapid first-phase decline appeared shorter (lasting only until day 2) in cultures treated with 250 U/ml, 500 U/ml, or 1,000 U/ml IFN-α (Fig. 2) and then was followed by a slower decline; however, since growth of the replicon cells treated with 500 U/ml or 1,000 U/ml IFN-α was compromised (i.e., slowed or stopped) by day 6 of treatment, the antiviral effect of IFN-α on HCV RNA levels cannot be accurately assessed after day 6 under these treatment conditions (i.e., some reduction in HCV RNA levels could be due to the decrease in cell growth). In Table 1 the sg1b RNA reduction rates (per day) and decline ratios from baseline for each IFN-α treatment are shown, indicating that the effect of IFN-α concentration on sg1b RNA levels was dose dependent (P < 0.01) (Fig. 2).

• Open in new tab
• Download powerpoint

FIG. 2.

Subgenomic HCV RNA kinetics during IFN-α treatment is biphasic and dose dependent. HCV replicon cells were incubated with various concentrations of IFN-α and harvested from triplicate wells at the indicated time points. Detailed quantification analysis is provided in Table 1. Asterisks indicate time points at which sg1b RNA levels were significantly different (P < 0.01) among the various IFN-α concentrations using a one-way analysis of variance test and posthoc (Bonferroni) test. The size of the symbols is larger than the standard error of the mean.

Detailed sg1b RNA inhibition kinetics during 100- and 250-U/ml IFN-α treatment.Because doses of 500 U/ml and 1,000 U/ml of IFN affect cell proliferation, which is known to negatively impact HCV RNA replication levels in cell culture (14, 23, 27), it may not be appropriate to analyze the kinetics of sg1b RNA decline under these conditions. However, to further characterize and dissect sg1b RNA IFN-α inhibition kinetics for the purposes of mathematical modeling, we repeated the inhibition assay at doses of 100 and 250 U/ml IFN-α but with more frequent sampling. Specifically, we harvested RNA from mock- and IFN-α-treated sg1b cells every 2 h during the first day, every 4 h during the second day, and then every 4 to 12 h until day 8 (Fig. 3).

• Open in new tab
• Download powerpoint

FIG. 3.

Kinetic analysis and modeling of the in vitro data for doses of 100 and 250 U/ml of IFN-α. HCV replicon cells were incubated with 100 or 250 U/ml of IFN-α, and triplicate RNA samples were collected for real-time qPCR analysis every 2 h from 0 to 24 h, every 4 h from 24 to 60 h, and then every 12 h until day 8. (A) RT-qPCR analysis of ISG56 mRNA levels in mock-treated cell (circles) and in cells treated with 100 U/ml (squares) and 250 U/ml (triangles) IFN-α throughout the 8-day experiment. (B) RT-qPCR analysis of sg1b HCV RNA levels in mock-treated cells (circles) and in cells treated with 100 U/ml (squares) and 250 U/ml (triangles) IFN-α during the first 24 h of treatment. Open arrow, sg1b RNA levels in IFN-α-treated cells significantly (P = 0.05) lower than in control cells; filled arrow, sg1b RNA levels in 250 U/ml IFN-α-treated cells significantly (P = 0.05) lower than in 100 U/ml IFN-α-treated cells. (C) Fit of model equations to sg1b RNA decline data. Solid line denotes the best fit of equations 2 and 4 or equation 6 with g of ∼0 to HCV RNA levels obtained with 100 U/ml IFN-α. Dashed line indicates the best fit of equation 6 to the measured 250-U/ml IFN-α sg1b RNA levels. The parameter values that generated the best-fit theoretical curves (solid and dashed lines) are shown in Table 2. The size of the squares represents the largest standard error of the mean of sg1b RNA. Vertical lines represent standard error of the mean.

To confirm constitutive, dose-dependent IFN signaling throughout the 8-day experiment, cellular expression levels of ISGs, ISG56 (Fig. 3A) and ISG15 (data not shown), were monitored. In Fig. 3A we show that ISG56 mRNA levels in IFN-α-treated cells is ∼1.5 logs higher than in the mock-treated cells by 6 h posttreatment and that this level is maintained throughout the experiment. Notably, by 12 h posttreatment ISG56 levels were constitutively higher in cells treated with 250 U/ml IFN-α than in cells treated with 100 U/ml IFN-α, directly demonstrating the dose responsiveness of the cells to the concentrations of IFN-α. A similar expression pattern was observed for ISG15 (data not shown).

Reminiscent of the delayed response observed in infected patients after the initiation of IFN-α therapy (16), during the first ∼12 h of IFN-α treatment the levels of HCV sg1b RNA in both the 100- and 250-U/ml IFN-α-treated cultures were similar to the level measured in the mock-treated control cultures (Fig. 3B). After this delay, the first-phase decline of sg1b RNA in the 100-U/ml IFN-α-treated cultures occurred at a rate of 0.79 ± 0.02 day⁻¹ ending at approximately day 4.5. This first-phase decline of 1.5 log₁₀ represents a 97% reduction in HCV sg1b RNA; hence, we define 100 U/ml IFN-α as having an in vitro efficacy of 97%. After this rapid 97% drop, a flat second phase was observed (characterized by a “decline” rate of 0.04 ± 0.07 day⁻¹, which is not significantly different from 0 [P = 0.6]), resulting in sg1b RNA levels that remained at ∼1.5 log₁₀ below baseline until day 8 (Fig. 3C and Table 1). In contrast, in the 250-U/ml cultures, a first-phase 1.5 log₁₀ decline of sg1b RNA occurred at a rate of 1.47 ± 0.07 day⁻¹, ending at approximately day 2.5. This decline rate is significantly (P < 0.001) higher than with 100 U/ml IFN-α, which can be noticed from 18 h postinitiation of treatment (Fig. 3B). After this rapid 97% drop, a slower decline ensued, resulting in sg1b RNA levels reaching 2.47 log₁₀ below baseline by day 8 (Fig. 3C and Table 1).

Notably, in both the 100- and 250-U/ml IFN-α-treated cultures, the majority of cells exhibited significantly reduced HCV NS5A protein levels by day 6, i.e., during the second phase of sg1b RNA decline (Fig. 4). However, it is interesting that a small percentage of cells (∼2%) in the IFN-α-treated cultures did retain slightly higher NS5A levels more comparable to the level observed in untreated cells (Fig. 4). Because these NS5A-positive cells did not disappear during treatment before the cells became confluent at day 8, it remains to be determined if HCV inhibition was randomly proceeding more slowly in these cells or if they represent a specific subset of cells that are more resistant to the effects of IFN-α. In either case, in the context of these experiments, the HCV RNA present in these cells does contribute to the number of HCV RNA copies present in the cultures throughout the experiment. Therefore, in the following sections we developed simple average or so-called mean-field models and fit them to the experimental data to provide insights into the viral RNA dynamics during IFN-α treatment.

• Open in new tab
• Download powerpoint

FIG. 4.

HCV NS5A levels during second-phase IFN-α inhibition of sg1b RNA. At 5, 6, 7, and 8 days posttreatment, mock-treated cells (A) and sg1b cell cultures treated with 100 U/ml (B) and 250 U/ml (C) IFN-α were fixed and stained for HCV NS5A. Cell nuclei were stained by Hoechst dye. Shown are representative ×20 images taken at day 6 posttreatment. HCV NS5A is shown in red (Alexa 555), and nuclei are shown in blue (Hoechst). Bar, 50 μm.

Modeling sg1b RNA inhibition kinetics during 100-U/ml IFN-α treatment.In order to model the effect of IFN-α on sg1b RNA, we initially focused on the 100-U/ml IFN-α inhibition data and assumed that in the absence of IFN-α, replicon cells produce sg1b RNA at a constant rate, α, and degrade sg1b RNA by a first-order process with rate constant μ. Notably, because replicon RNA is normalized to the total cellular RNA, reduction of sg1b RNA copy number on a per-cell basis due to cell division can be neglected, assuming that total cellular RNA and sg1b RNA are diluted to the same extent as replicon cells divide.

In order to account for the reduction of sg1b RNA levels during IFN-α therapy, we tested two alternative hypotheses. First, we assumed some process that supports sg1b RNA production after a time (t) delay, t₀, is affected by IFN-α, resulting in the rate of sg1b RNA production changing from α to (1 − ε_in)α, where ε_in (0 ≤ ε_in ≤ 1) is the effectiveness of IFN-α in reducing intracellular sg1b RNA production. Thus, if R is the intracellular concentration of sg1b RNA, this model can be described by the equation $$mathtex$$\[\frac{dR}{dt}{=}(1{-}{\varepsilon}_{\mathrm{in}}){\alpha}{-}{\mu}R\]$$mathtex$$(1) with the solution $$mathtex$$\[R(t){=}R(0)[1{-}{\varepsilon}_{\mathrm{in}}{+}{\varepsilon}_{\mathrm{in}}\ \mathrm{exp}({-}{\mu}(t{-}t_{0}))]\]$$mathtex$$(2) where t₀ is a parameter introduced to account for the delay from the time of IFN-α administration until sg1b RNA decline is observed, R(0) is the amount of intracellular sg1b RNA at the start of therapy, i.e., at t = 0, and we have assumed that at the start of therapy α = μR(0), i.e., replicon production has reached steady state. We make this assumption because the level of sg1b RNA is constant in control cells and because the level in IFN-α-treated cells is also approximately constant for the first 12 h after IFN-α administration (Fig. 3B). Fitting the natural logarithm of R(t), given by equation 2 to the natural logarithm of the experimental data using nonlinear least squares regression, we find that there is good agreement between the model and the data (Fig. 3C), with the best-fit parameter values being a ε_in of 0.972, μ of 0.98 day⁻¹, and t₀ of 9.5 h (Table 2). Hence, under the assumption that only the sg1b RNA production rate is being altered during IFN-α treatment, parameter estimates suggest that 100 U/ml of IFN can decrease sg1b RNA production by 97.2% and that at steady state sg1b RNA is degraded by a first-order process with a rate constant of ∼0.98 day⁻¹, which corresponds to an sg1b RNA half-life (t_1/2) of (ln 2)/0.98, or 0.71 days, i.e., 17 h (95% confidence interval [CI], 16 to 19 h). This t_1/2 estimate is somewhat shorter but corresponds reasonably well with the 19 h (95% CI, 18 to 21) we calculated by linear regression during 4 days of treatment of the sg1b cells with the HCV NS5B inhibitor, NM107, which specifically blocks HCV RNA production (N. Baretto et al., unpublished data). Likewise, the 9.5-h estimated value of t₀ calculated by equation 2 corresponds well with the empirically observed delay in the sg1b RNA decline we observed after IFN-α is added to the culture medium.

Alternatively, we examined the hypothesis that instead of reducing RNA production, IFN-α might increase replicon clearance/degradation. To describe this scenario, the following alternative form of the model is used: $$mathtex$$\[\frac{dR}{dt}{=}{\alpha}{-}\frac{{\mu}}{(1{-}{\varepsilon}_{\mathrm{in}})}R\]$$mathtex$$(3) where 1/(1 − ε_in) is the factor by which clearance is increased. Since ε_in is <1, this factor is greater than 1. If one again assumes the system is at steady state before IFN-α treatment, the solution of this model is $$mathtex$$\[R(t){=}R(0)\ \left[1{-}{\varepsilon}_{\mathrm{in}}{+}{\varepsilon}_{\mathrm{in}}\ \mathrm{exp}\left({-}\frac{{\mu}}{1{-}{\varepsilon}_{\mathrm{in}}}(t{-}t_{0})\right)\right]\]$$mathtex$$(4) Note that this equation is the same as equation 2 except that μ in equation 2 is replaced by the expression μ̂ = μ/(1 − ε_in) in equation 4. However, if we use equation 4 to fit the experimental data, we obtain the parameter estimates of a ε_in of 0.972, μ of 0.027 day⁻¹, and t₀ of 9.5 h (Table 2). This smaller estimate for the natural rate of loss of replicons, μ, occurs because we have assumed that in the presence of IFN-α the loss rate increases by the factor 1/(1 − ε_in), i.e., ∼30-fold. Thus, although this model gives exactly the same fit to the data as the one shown in Fig. 3C, it predicts that the intracellular t_1/2 of replicon RNA in the absence of IFN-α is (ln 2)/0.027, or 26 days.

Given the usual rapid turnover of RNAs in cells and the sg1b t_1/2 of 19 h measured in the presence of HCV NS5B inhibitor (Baretto et al., unpublished), we can therefore reject the latter hypothesis in favor of the former, which states that IFN-α is acts by reducing the level of HCV RNA production. While it is still possible that IFN-α not only reduces sg1b RNA production but also has a smaller effect on increasing sg1b RNA clearance, this analysis supports the conclusion that inhibition of HCV RNA synthesis is the primary effect observed during treatment with 100 U/ml IFN-α.

Modeling sg1b RNA inhibition kinetics during 250-U/ml IFN-α treatment.Examining equation 1 suggests that, rather than having a flat second phase, a decline of sg1b RNA during treatment with 250 U/ml IFN would occur if the intrinsic rate of replicon production, α, in addition to being initially inhibited by a factor of 1 − ε_in, also decreases with time on therapy, say due to the net destruction of negative-strand RNA or the gradual loss of some other components of the replication complex. The continued reduction of such replication components would be expected to result in a further slowing of HCV sg1b RNA synthesis after the completion of the first phase, resulting in a decrease in sg1b RNA levels in the second phase. To incorporate this concept into the model, we assume that (1 − ε_in)α, rather than being a constant, is a function that decreases with time, for example, (1 −ε_in)α₀ exp(−gt), where g is a constant and t is time posttreatment. Then equation 1 would become $$mathtex$$\[\frac{dR}{dt}{=}(1{-}{\varepsilon}_{\mathrm{in}}){\alpha}_{0}\ \mathrm{exp}({-}gt){-}{\mu}R\]$$mathtex$$(5) with the solution $$mathtex$$\[R(t){=}R(0)\ \left[\frac{(1{-}{\varepsilon}_{\mathrm{in}}){\mu}}{{\mu}{-}g}\ \mathrm{exp}({-}g(t{-}t_{0})){+}\frac{{\varepsilon}_{\mathrm{in}}{\mu}{-}g}{{\mu}{-}g}\ \mathrm{exp}({-}{\mu}(t{-}t_{0}))\right]\]$$mathtex$$(6) where we have assumed a delay, t₀, before IFN-α has any effect and that replicon production is at steady state at the start of therapy, i.e., α₀ = μR(0). Importantly, comparing equation 6 to our previous simple model, equation 2, shows that when g is 0, the two equations are identical, and the same flat second phase predicted in Fig. 3C (solid line) is obtained. However, as intended, for any g that is >0, the flat second phase disappears, and we instead obtain a decline in the second phase (Fig. 3C). Fitting the natural logarithm of R(t), given by equation 6 to the natural logarithm of the 250-U/ml IFN-α experimental data (Fig. 3C) using nonlinear least squares regression, we find that there is good agreement between the model and the data, with the best-fit parameter values being a ε_in of 0.927, μ of 2.32 day⁻¹, g of 0.33 day⁻¹, and t₀ of 7.8 h (Table 2). Interestingly, while the estimated degrees of inhibition of the sg1b RNA production rate under 100 and 250 U/ml IFN-α, ε_in, have overlapping 95% CIs (Table 2), the sg1b RNA degradation rate, μ, is approximately twofold higher in the presence of 250 U/ml.