T-bet⁺ Treg Cells Undergo Abortive Th1 Cell Differentiation due to Impaired Expression of IL-12 Receptor β2

STAT1-Dependent Expression of T-bet by Treg Cells

We recently identified and characterized a population of T-bet⁺CXCR3⁺ Treg cells that express elevated amounts of Il10 and that are important for regulating strong Th1 cell responses (Koch et al., 2009). Moreover, the frequency of this Treg cell population is increased during Th1 cell responses, but substantially reduced in mice lacking STAT1. This suggests that like effector T cells, Treg cells upregulate T-bet during Th1 cell responses after STAT1 activation. Consistent with this, nearly all Foxp3⁺ Treg cells rapidly phosphorylated STAT1 after treatment with the STAT1-activating cytokines IFN-γ or IL-27 (Figure S1 available online). Additionally, in vitro stimulation in the presence of either IFN-γ or IL-27 potently induced T-bet expression in sorted CD4⁺Foxp3⁺CD62L⁺CXCR3⁻ naive phenotype Treg cells (Figure 1 A). To determine whether Treg cells are exposed to STAT1-activating cytokines in vivo, we constructed mixed bone marrow (BM) chimeras using congenically marked wild-type and Stat1 ^−/− donors and compared the phenotype of Foxp3⁺ Treg cells derived from each donor 8 weeks later. Indeed, whereas ∼25%–30% of the splenic Treg cells from the wild-type donor were T-bet⁺CXCR3⁺, this population of Treg cells was nearly absent from the Stat1 ^−/− cells (Figure 1B).

Treg cell expression of T-bet and CXCR3 is reduced in mice lacking the STAT1-activating IFN-γ receptor (IFN-γR [Koch et al., 2009]), indicating that IFN-γ is the principle STAT1-activating cytokine driving the development of T-bet⁺CXCR3⁺ Treg cells. Moreover, CXCR3⁺ Treg cells from Ifngr ^−/− mice displayed a slight (<2-fold) decrease in T-bet protein expression relative to CXCR3⁺ Treg cells from wild-type animals (Figure S2), further highlighting the importance of IFN-γR signaling for optimal T-bet induction in Treg cells. However, tissue inflammation results in the upregulation of numerous inflammatory and regulatory cytokines, several of which (i.e., IL-27) can activate STAT1 and may induce T-bet and CXCR3 expression in Treg cells in the absence of IFN-γ. To determine whether these cytokines can compensate for loss of IFN-γ signaling during strong inflammatory responses in vivo, we induced experimental autoimmune encephalomyelitis (EAE), a demyelinating inflammatory disease mediated by autoreactive Th1 and Th17 cells (Dardalhon et al., 2008), in mixed BM chimeras containing a 1:1 ratio of congenically marked wild-type and Ifngr1 ^−/− cells. Loss of IFN-γR-mediated signaling did not impair the accumulation of CD4⁺ T cells in the central nervous system (CNS; data not shown). However, expression of T-bet and CXCR3 were substantially reduced in Ifngr1 ^−/− Treg cells in both the CNS and spleen (Figure 1C and data not shown). Additionally, consistent with the large amount of Il10 expressed by CXCR3⁺ Treg cells, Ifngr1 ^−/− Treg cells in the CNS showed decreased IL-10 production at day 19 (Figure 1D), a late time point associated with Treg cell-mediated amelioration of disease (Korn et al., 2007; McGeachy et al., 2005). Thus, even during strong inflammatory responses associated with EAE, IFN-γ promotes the functional specialization of Treg cells by controlling the expression of molecules such as CXCR3 and IL-10 that are critical for Treg cell-mediated disease suppression (McGeachy et al., 2005; Müller et al., 2007).

T Cell Derived-IFN-γ Drives T-bet⁺ Treg Cell Differentiation

The homeostasis of Treg cells is controlled in part by IL-2 produced by CD4⁺CD44^hiFoxp3⁻ effector T cells (Setoguchi et al., 2005), establishing a link between effector T cell activation and Treg cell abundance. To determine whether effector T cells also control Treg cell specialization via production of IFN-γ, we injected a 9:1 ratio of BM cells from Tcrb ^−/− Tcrd ^−/− and Ifng ^−/− donors into irradiated Tcrb ^−/− Tcrd ^−/− recipients in order to construct animals in which all nonhematopoietic cells, as well as 90% of non-T cells, were IFN-γ sufficient, whereas all T cells were derived from Ifng ^−/− donor BM. Control chimeras were generated using a 9:1 mixture of Tcrb ^−/− Tcrd ^−/− and wild-type BM cells. We found the that frequency of T-bet⁺CXCR3⁺ Treg cells in the spleens, peripheral lymph nodes, and peritoneal cavities of mice lacking T cell-derived IFN-γ was substantially lower than in the wild-type control chimeras (Figure 2 A).

To determine whether IFN-γ produced by effector T cells is sufficient to induce CXCR3 and T-bet expression by Treg cells, we transferred OVA-specific CD4⁺ Th1 cells from OT-II TCR transgenic mice into wild-type and Ifng ^−/− recipients. These mice were then given OVA_323-339 peptide (OVAp) in PBS on days 1 and 4 after transfer, and we examined the endogenous Treg cells in these animals 3 days later. Indeed, in both wild-type and Ifng ^−/− mice given OT-II Th1 cells and OVAp, we observed a 3- to 4-fold increase in the fraction of T-bet⁺CXCR3⁺ Treg cells as compared to mice receiving OVAp alone (Figure 2B). Similarly, activation of effector CD8⁺ T cells from OT-I TCR transgenic mice in vivo also increased Treg cell expression of T-bet and CXCR3 (Figure S3). Thus, through the production of IFN-γ, both CD4⁺ and CD8⁺ effector T cells can directly control the specialization of Treg cells.

IL-12R Signaling Is Dispensable for the Development of T-bet⁺ Treg Cells

In addition to the STAT1 activating cytokines, IL-12 has an essential role in Th1 cell differentiation (Trinchieri, 2003; Hsieh et al., 1993; Manetti et al., 1993), and activation of STAT4 via the IL-12 receptor is necessary for expression of genes required for Th1 cell function (Thieu et al., 2008; Wei et al., 2010). However, Treg cell expression of T-bet and CXCR3 was unaltered in IL-12Rβ2-deficient mice (Figures 3 A and S2), indicating that STAT4 activation via IL-12R is dispensable for the differentiation of T-bet⁺ Treg cells.

Although most T cells express the IL-12 receptor subunit IL-12Rβ1, naive CD4⁺ T cells lack expression of IL-12Rβ2 and are not IL-12 responsive (Szabo et al., 1997). Induction of IL-12Rβ2 downstream of STAT1 and T-bet is a key checkpoint in Th1 cell development (Afkarian et al., 2002; Mullen et al., 2001). Indeed, whereas both Foxp3⁺CXCR3⁺ Treg cells and Foxp3⁻CXCR3⁺ Th1 cells expressed Il12rb1, expression of Il12rb2 was ∼100-fold lower in CXCR3⁺ Treg cells (Figure 3B). Accordingly, these cells failed to phosphorylate STAT4 in response to IL-12 in vitro (Figure 3C). To determine whether Foxp3⁺CXCR3⁺ Treg cells showed evidence of STAT4 activation, we compared expression of several STAT4-controlled genes in sorted Foxp3⁺CXCR3⁺ Treg cells and Foxp3⁻CXCR3⁺ Th1 cells (Wei et al., 2010). Consistent with their lack of IL-12 responsiveness, expression of each of these genes was substantially lower in the CXCR3⁺ Treg cells compared with the Foxp3⁻ Th1 cells (Figure 3D). Thus, despite having undergone STAT1-dependent T-bet upregulation, CXCR3⁺ Treg cells failed to acquire IL-12 responsiveness through upregulation of IL-12Rβ2 and consequently lacked the STAT4-dependent genetic signature characteristic of fully differentiated Th1 cells.

Differential Regulation of IL-12Rβ2 Expression in Foxp3⁺ and Foxp3⁻ Cells

The striking lack of IL-12 responsiveness in T-bet⁺CXCR3⁺ Treg cells suggests that regulation of IL-12Rβ2 expression is different in Foxp3⁺ and Foxp3⁻ cells. Alternatively, chronic TCR stimulation can inhibit IL-12Rβ2 induction (Schulz et al., 2009), and this may limit its expression in autoreactive Treg cells. To differentiate between these possibilities, we compared acquisition of IL-12 responsiveness in Foxp3⁺ and Foxp3⁻ CD4⁺ T cells stimulated in vitro under identical conditions. Sorted naive phenotype CD4⁺CD62L⁺CXCR3⁻Foxp3⁻ and CD4⁺CD62L⁺CXCR3⁻Foxp3⁺ T cells were activated in the presence of IL-2 and IFN-γ for 3 days, at which point the cells were removed from the TCR stimulus and cultured in cytokines for three additional days. We measured STAT4 phosphorylation after IL-12 treatment in the Foxp3⁺ and Foxp3⁻ cells daily starting at day 2 of culture. IL-12 responsiveness in the Foxp3⁻ cells was evident by day 2 of activation, peaked by day 4, and was maintained throughout the remainder of the culture (Figure 4 A, top). In comparison, acquisition of IL-12 responsiveness was substantially delayed in the Foxp3⁺ Treg cells, which did not become IL-12 responsive until day 4 of culture (Figure 4A, bottom). However, by day 6 the Foxp3⁺ and Foxp3⁻ cells displayed roughly equivalent STAT4 phosphorylation after IL-12 treatment. Importantly, Treg cells became IL-12 responsive on day 4 of culture even when removed from the TCR stimulus at earlier time points (day 1 or 2) postactivation (data not shown), indicating that TCR stimulation is not the primary factor inhibiting their expression of Il12rb2.

The delay in acquisition of IL-12 responsiveness suggests that T-bet-dependent induction of Il12rb2 expression is impaired in Treg cells. To directly test this, we sorted Foxp3⁺ and Foxp3⁻ CD4⁺T cells from T-bet-deficient Tbx21 ^−/−Foxp3^gfp mice, stimulated them in vitro with plate-bound anti-CD3 and anti-CD28, and infected them with a retroviral construct encoding a T-bet-estrogen receptor (T-bet-ER) fusion protein whose activity is modulated by the estrogen analog 4-hydroxytamoxifin (4-HT). The infected cells were cultured for 3 days with varying concentrations of 4-HT, at which point we assessed their expression of Cxcr3 and Il12rb2. As expected, activation of T-bet led to robust induction of Cxcr3 mRNA expression in both Foxp3⁺ and Foxp3⁻ cells (Figure S4). However, induction of Il12rb2 was observed only in Foxp3⁻ cells (Figure 4B), demonstrating that the regulation of this gene is markedly different in Foxp3⁺ and Foxp3⁻ CD4⁺ T cells.

To explore the molecular basis for the differential regulation of Il12rb2 in Foxp3⁺ and Foxp3⁻ cells, we compared the epigenetic status of the Il12rb2 promoter in naive CD4⁺ T cells and Treg cells through analysis of genome-wide histone modification data (Wei et al., 2009). In naive CD4⁺ T cells, the Il12rb2 promoter is marked by both permissive H3K4me3 and repressive H3K27me3 histone modifications (Figure 4C, top). Consistent with the rapid induction of Il12rb2 in developing Th1 cells, this bivalent state is believed to repress gene transcription while keeping the promoter “poised” for activation-induced expression (Bernstein et al., 2006). By contrast, only H3K27me3 marks are found at the Il12rb2 promoter in Treg cells (Figure 4C, bottom). Thus, expression in Treg cells probably requires substantial epigenetic remodeling of the Il12rb2 promoter, involving both the removal of repressive histone modifications and the addition of permissive modifications.

IL-12 Can Functionally Reprogram Treg Cells

Although IL-12 has been reported to induce IFN-γ-production by Foxp3⁺ Treg cells in vitro, it does so inefficiently, with only ∼5% of CD4⁺Foxp3⁺ cells producing IFN-γ after activation for 3 days in the presence of IL-12 (Wei et al., 2009). Moreover, a recent study demonstrating that Foxp3⁺ cells contain a small population of effector T cells suggests that at least some of the observed IFN-γ production could result from these contaminating cells (Miyao et al., 2012). Indeed, we found that IL-12 alone failed to induce IFN-γ expression by sorted CD4⁺CD62L⁺CXCR3⁻Foxp3⁺ naive phenotype Treg cells, which are less likely to contain contaminating activated effector cells (Figure S5).

The stringent regulation of IL-12Rβ2 expression in autoreactive Treg cells may serve as a tolerance mechanism that prevents them from undergoing IL-12-dependent functional differentiation and acquiring potentially dangerous proinflammatory properties such as the ability to produce IFN-γ. Alternatively, the inability of Treg cells to produce IFN-γ may result from direct inhibition of Ifng expression by Foxp3 or other Treg cell-specific repressive mechanisms. To determine how IL-12 receptor signaling functionally impacts Treg cells, we activated sorted CD4⁺Foxp3⁺CD62L⁺CXCR3⁻ naive phenotype Treg cells in the presence of IFN-γ for 6 days to render the cells IL-12 responsive. We then cultured these cells with or without IL-12 for 3 additional days, examined their ability to produce IFN-γ, and assessed their suppressive function in vitro. Despite the reported ability of Foxp3 to directly inhibit IFN-γ expression (Ono et al., 2007), IL-12 potently induced IFN-γ production in both Foxp3⁺ and Foxp3⁻ cells that had been “primed” with IFN-γ (Figure 5 A). In addition, IL-12 treatment substantially reduced the ability of Treg cells to inhibit effector T cell proliferation (Figure 5B). Thus, IL-12 can alter the balance between pro- and anti-inflammatory functions in Treg cells, emphasizing the importance of regulating IL-12 responsiveness in order to maintain Treg cell-mediated immune suppression during strong type 1 inflammatory responses.

Treg Cells Undergo Abortive Th1 Cell Development during Acute Bacterial Infection

Although robust, production of IL-12 during infection with intracellular pathogens is typically very transient. Because Treg cells are poised to respond to IFN-γ, but display delayed induction of IL-12Rβ2, we predicted that during acute type 1 inflammatory responses they would undergo partial Th1 cell differentiation driven by activation of STAT1 but not STAT4. To test this, we monitored cytokine responses in different splenic lymphocyte populations throughout the course of acute infection with the intracellular bacterial pathogen L. monocytogenes, which provokes a strong, IL-12-dependent Th1 response.

Consistent with the transient appearance of IL-12p70 in the serum of L. monocytogenes-infected mice (Way et al., 2007; Seki et al., 2002), the Il12a and Il12b genes were strongly upregulated in the spleen within 1 day of infection, after which their expression was rapidly extinguished (Figure 6 A and data not shown). Moreover, direct ex vivo analysis of CD4⁺Foxp3⁻CD44^hi T cells (without any in vitro restimulation) showed that ∼25% of these effector T cells in the spleen were phospho(p)-STAT4⁺ on day 1 after infection and that a substantial fraction of the pSTAT4⁺ cells were also producing IFN-γ (Figures 6B and 6C). However, mirroring the sharp decline in IL-12 cytokine expression, the frequency of pSTAT4⁺ IFN-γ-producing effector T cells fell dramatically on days 2 and 3 after infection. By contrast, less than 5% of splenic Treg cells were pSTAT4⁺ directly ex vivo at any time point examined during infection (Figures 6B, middle, and 6C). Indeed, IL-12 responsiveness, as measured by the ability to phosphorylate STAT4 after in vitro IL-12 treatment, emerged in only a small fraction of Treg cells between days 3–5 postinfection, after production of IL-12 had ceased (Figure 6B, right).

Coincident with the robust IFN-γ production by effector T cells, we observed a dramatic increase in STAT1 phosphorylation in Treg cells on days 1 and 2 postinfection (Figure 7 A). Accordingly, expression of T-bet and CXCR3 was dramatically elevated in Treg cells from L. monocytogenes infected mice. However, consistent with their inability to respond to IL-12 early in infection, very few Treg cells acquired the ability to produce IFN-γ (Figure 7B). To determine whether IL-12 receptor signaling is required for the small degree of IFN-γ production by Treg cells during L. monocytogenes infection, we reconstituted Tcrb ^−/− Tcrd ^−/− mice with a 1:1 mixture of splenocytes from wild-type and Il12rb2 ^−/− donors and infected them with L. monocytogenes 10 days later. Indeed, whereas a small fraction of wild-type Treg cells acquired the ability to produce IFN-γ after infection, this was blunted in the Il12rb2 ^−/− Treg cells (Figure 7C). Taken together, these data confirm our in vitro findings regarding the ability of Treg cells to respond to IFN-γ and IL-12 in an in vivo setting of acute inflammation and indicate that the relative inability of Treg cells to respond to IL-12 limits their functional reprogramming into IFN-γ-producing cells.

Mice

C57BL/6J, B6.129P2-Tcrb ^tm1Mom Tcrd ^tm1Mom/J (Tcrb ^−/− Tcrd ^−/−), B6.129S7-Ifngr1 ^tm1Agt /J (Ifngr1 ^−/−), B6.129S7-Ifng ^tm1Ts /J (Ifng ^−/−), B6.129S1-Il12rb2 ^tm1Jm/J (Il12rb2 ^−/−), B6.Cg-Tg(TcraTcrb)425Cbn/J (OT-II), and B6-Tg(TcraTcrb)1100Mjb/J (OT-I) mice were purchased from the Jackson Laboratory. B6.SJL-Ptprc ^a/BoyAiTac (CD45.1) mice were purchased from Taconic Farms. Foxp3^gfp mice were provided by A. Rudensky (Memorial Sloan-Kettering Cancer Center). Tissues from Stat1 ^−/− mice were provided by M. Krishna-Kaja (University of Washington). All animals were housed and bred at the Benaroya Research Institute Seattle, WA and experiments were performed in accordance within the guidelines of the Animal Care and Use Committees of the Benaroya Research Institute.

Cell Isolations

Lymphocytes were isolated from the spleens and lymph nodes (LNs) by tissue disruption with glass slides followed by filtration through a 100 μM filter. CNS lymphocytes were recovered from perfused animals by grinding brain and spinal cord tissue through a 100 μm filter with the plunger of a 5 mL syringe. Cells were pelleted by centrifugation at 3200 rpm for 15 min and resuspended in 6 mL of 37% Percoll (GE Healthcare). Cells were centrifuged again at 3200 rpm for 15 min (brake off) and lymphocytes were recovered after evacuation of the fatty supernatant.

Bone Marrow Chimeras

BM cells were depleted of CD4⁺ cells with anti-CD4 microbeads (Miltenyi Biotech) and injected retro-orbitally into lethally irradiated (1000 Rad) Tcrb^−/−Tcrd^−/− mice. Chimeras received 6 to 10 × 10⁶ cells of a 1:1 mixture of wild-type (CD45.1⁺) and knockout (CD45.2⁺) BM, or 9 × 10⁶ TCRβδ-KO BM cells mixed with 1 × 10⁶ wild-type or Ifng ^−/− BM cells as indicated.

Flow Cytometry and Cell Sorting

Data were acquired on LSRII flow cytometers (BD Biosciences) and analyzed with FlowJo software (Treestar). For cell sorting experiments, CD4⁺ cells were enriched with Dynal CD4 T cell negative isolation kit (Invitrogen), stained for desired cell surface markers, and isolated with a FACS Vantage or FACS Aria (BD Biosciences). For intracellular staining of IFN-γ, Foxp3, and/or T-bet, lymphocytes were surface stained, then permeabilized with the eBioscience FixPerm buffer. For intracellular cytokine staining after restimulation, cells were stimulated with 50 ng/mL phorbol 12-myristate 13-acetate (PMA) and 1 μg/mL ionomycin in the presence of 10 μg/mL monensin for 4 hr at 37°C, 5%CO₂ prior to staining.

Phospho-STAT Staining

Cells were stimulated for 30 min in the presence or absence of 25 ng/mL recombinant mouse (rm)IL-12 (eBioscience), IFN-γ (Biolegend) or IL-27 (eBioscience). Cells were then and fixed for 20 min in BD Fix and Perm buffer at room temperature (BD Biosciences), washed with BD Perm Wash buffer, and fixed in 90% ice-cold methanol for 30min on ice. Cells were washed two times with BD Perm/Wash and stained with antibodies against cell surface and intracellular markers, including pSTAT4 (Y693; BD Biosciences) and pSTAT1 (Y701; BD Biosciences), in BD Perm Wash for 45 min at room temperature. For direct ex vivo pSTAT and cytokine staining, ∼1/5 of each spleen was ground between glass slides in BD Fix and Perm buffer, left for 20 min at room temperature, washed, fixed in 90% methanol, and stained as described above.

In Vitro T Cell Culture

CD4⁺CD62L⁺CXCR3⁻ Foxp3⁺ and Foxp3⁻ cells were FACS sorted from Foxp3^gfp mice. A total of 1.5 to 2 × 10⁶ cells were stimulated with plate-bound anti-CD3 (2C11; UCSF hybridoma core) and anti-CD28 (37.51; UCSF hybridoma core) in medium containing 500 units/mL rmIL-2. rmIFN-γ, rmIL-27, and rmIL-12p70 were added to some cultures at 25 ng/mL as indicated. All cytokines used were from eBioscience or Biolegend. Cells were cultured for 72 hr at 37°C, at which point they were analyzed or removed from the CD3 and CD28 stimulus and cultured in fresh medium containing cytokines for 48–72 more hours at 37°C. In some assays, after 144 hr of culture, cells were washed and cultured for an additional 3 days with 500 units/mL IL-2 in the presence or absence of 25 ng/mL IL-12.

Adoptive Transfer and OVA Peptide Administration

A total of 5 × 10⁶ polarized OT-II⁺CD4⁺ or OT-I⁺CD8⁺ cells were adoptively transferred into wild-type or Ifng ^−/− mice via retro-orbital injection on day 0 and recipients were injected intraperitoneally with 200 μg OVA peptide_323-339 (for OT-II adoptive transfers) or OVA peptide_257-264 (for OT-I adoptive transfers) in PBS on days 1 and 4. All mice were sacrificed for analysis on day 7.

In Vitro Suppression Assays

A total of 7.5 × 10⁵ CFSE-labeled CD4⁺CD25⁻ T cells were incubated with 7.5 × 10⁵ irradiated (2,500 Rad) Tcrb ^−/− Tcrd ^−/− splenocytes (CD45.2⁺), with or without addition of Treg cells at the indicated ratios, and stimulated with 0.15 μg/ml soluble anti-CD3 (2C11) for 72 hr.

L. Monocytogenes Infection

The recombinant LM-OVA strain was provided by M. Bevan (University of Washington). A total of 10⁴ cfu LM-OVA were injected intravenously into mice after growth to log phase (OD₆₀₀ = 0.1–0.4) in brain heart infusion media (BD Biosciences) at 37°C.

Induction of EAE

Mice were immunized subcutaneously (s.c.) in the flank with 200 μg myelin oligodendrocyte glycoprotein (MOG) 35–55 peptide in CFA. A total of 200 ng of pertussis toxin in PBS was injected intravenously on day 0 and day 2 after immunization.

Retroviral Transduction Assays

FACS-purified CD4⁺Foxp3⁺ and CD4⁺Foxp3⁻ cells from Tbx21 ^−/− Foxp3^gfp mice were plated in a 24-well flat-bottom plate precoated with 1 μg/ml anti-CD3 (2C11) and 1 μg/ml anti-CD28 (37.51) in medium with 500 units/ml rmIL-2. After 72 hr, cells were resuspended in retroviral supernatant containing 8 μg/ml polybrene (Sigma-Aldrich) and were centrifuged at 2500 rpm for 2 hr at 37°C, then cultured at 37°C in complete RPMI containing 500 U/ml rmIL-2 and 4-hydroxytamoxifen (Sigma-Aldrich). Infected cells were isolated 72 hr later with anti-human CD2 microbeads (Miltenyi Biotech) and expression of Cxcr3 and Il12rb2 was assessed by quantitative PCR.

Quantitative PCR

RNA extraction was performed with QIAGEN RNeasy columns (QIAGEN) and cDNA was generated with an Omniscript RT Kit (QIAGEN) according to the manufacturer’s instructions. Presynthesized Taqman Gene Expression Assays (Applied Biosystems) were used for amplifying Il12Rb2 (Mm00434200_m1) or Cxcr3 (Mm99999054_s1) mRNA transcripts in analysis of T-bet-ER transduced cells. Target gene values are expressed relative to Actb, which was analyzed with the sense primer 5′-TGACAGGATGCAGAAGGAGAT-3′, anti-sense primer 5′-GCGCTCAGGAGGAGCAAT-3′, and probe 5′-FAM-ACTGCTCTGGCTCCTAGCACCAT-TAMRA-3′. In all other assays, expression of Il12rb1, Il12rb2, Il12a, Nkg7, Il21, Id2, and Ffar were assessed with Maxima SYBR Green/ROX qPCR Master Mix (Fermentas) and normalized to expression of Gapdh.