Selenium metabolism in mammals

The possible means of how selenium is metabolized in mammalian systems is shown in Fig. 1. Se can potentially make its way into protein by multiple pathways, but only one of these pathways is used to specifically incorporate this element into selenoproteins: through the Se-containing amino acid, selenocysteine (Sec), the 21st amino acid in the genetic code⁽Reference Hatfield and Gladyshev¹⁹⁾. Sec is programmed to be inserted into protein by the stop codon, UGA, and thus, this codeword serves as a dual function in mammals. Its biosynthesis occurs by a novel pathway in eukaryotes in that the synthesis is carried out on Sec tRNA, designated Sec tRNA^[Ser]Sec and proceeds as follows: (1) serine is attached to tRNA^[Ser]Sec in the presence of seryl-tRNA synthetase and ATP to form Ser-tRNA^[Ser]Sec; (2) the seryl moiety is phosphorylated in the presence of O-phosphoseryl-tRNA^[Ser]Sec kinase and ATP to form O-phosphoseryl-tRNA^[Ser]Sec and (3) O-phosphoseryl-tRNA^[Ser]Sec in turn serves as a substrate for Sec synthase. In this reaction, monoselenophosphate, which is the active donor of Se, donates Se to the acceptor intermediate generated by Sec synthase to yield selenocysteyl-tRNA^[Ser]Sec. Monoselenophosphate is synthesized by selenophosphate synthetase 2 from selenide (or other Se compounds) and ATP. The details of Sec biosynthesis have been reviewed elsewhere⁽Reference Xu, Carlson and Zhang²⁰⁾. The other pathways wherein Se can be non-specifically incorporated into protein, namely by selenomethionine replacing methionine and Sec replacing cysteine, may result in deleterious effects to the cell, especially if such insertion occurs to any appreciable extent.

Fig. 1. Se and selenomethionine metabolism in mammals. Pathways of Se utilization in mammals are shown. Further discussion of Se and selenomethionine metabolism may be found elsewhere⁽Reference Finley^40, Reference Jackson and Combs⁴¹⁾. Sec, selenocysteine (see text); Cys, cysteine.

mRNA expression and T-cell development

There are 24 selenoprotein genes in mice⁽Reference Kryukov, Castellano and Novoselov²⁴⁾ and individual selenoproteins and selenoprotein mRNA manifest tissue specificity in their patterns of expression and in the manner in which their expression is regulated⁽Reference Hoffmann, Hoge and Li²⁵⁾. We examined the expression of the mRNA levels transcribed from each of the 24 selenoprotein genes by isolating total RNA from the three lymphoid organs, lymph nodes, spleen and thymus of normal mice and analysing the expression by real-time PCR. As shown in Fig. 2(A), GPx1 mRNA was highly expressed in all three organs. Several other selenoprotein mRNA, namely Sep15, GPx4, SelP, SelK, SelT and SelH, were moderately-to-highly expressed in each organ. Several additional selenoprotein mRNA were also made, while others were only slightly, or not at all, detectable. It is not clear why there is some variation in the levels of selenoprotein mRNA expression between the three tissues.

Fig. 2. Selenoprotein expression in T-cells. (A) Expression of selenoprotein mRNA was assessed by quantitative real-time PCR. Purified T-cells were obtained from lymph nodes, spleen and thymus of mice as described⁽Reference Shrimali, Irons and Carlson²¹⁾. (B) Expression of selenoproteins was assessed by ⁷⁵Se-labelling as described⁽Reference Carlson, Novoselov and Kumaraswamy²⁷⁾. The identities of major labelled selenoproteins are designated on the right panel; and the left panel shows Coomassie Blue-stained (CBB) gel that was used as a loading control.

We also examined selenoprotein expression in T-cells by labelling control and selenoprotein deficient (ΔTrsp^T) mice with ⁷⁵Se, isolating the T-cells from the thymus and examining the labelled selenoproteins by gel electrophoresis (Fig. 2(B)). Several selenoproteins were labelled in control mice, some of which have been identified in earlier studies⁽Reference Gladyshev, Stadtman and Hatfield²⁶⁾. Virtually no selenoproteins were labelled in the ΔTrsp^T T-cells and the faint labelling is most likely due to contamination from other cell types⁽Reference Carlson, Novoselov and Kumaraswamy²⁷⁾.

The thymus, spleen and lymph nodes of the affected mice manifested moderate-to-severe atrophy wherein their mass and cellularity were decreased to varying degrees ranging between 50% and 80% of those of control mice. The T-cell population in lymphoid organs, which was analysed by flow cytometry, was altered significantly. For example, the CD3⁺ population in splenocytes of ΔTrsp^T mice was about 50% of that of control splenocytes, while the CD8⁺ T-cell sub-population was more substantially reduced (>50%). Although the consequences of the partial loss of functional T-cells and in particular CD8⁺ T-cells, was not clearly defined in our study, other studies have shown that Se deficiency in the diet, which is known to cause a loss or reduction in selenoprotein function, impairs antiviral immunity and exacerbates viral pathogenesis⁽Reference Beck^28, Reference Hurwitz, Klaus and Llabre²⁹⁾.

T-cell activation

We next tested the ability of selenoprotein-deficient T-cells to be activated by T-cell receptor (TCR) signalling. Purified control T-cells and ΔTrsp^T T-cells were cultured in the presence or absence of anti-CD28 and anti-CD3 antibodies wherein the presence of these antibodies mimicked TCR activation and stimulated T-cells to proliferate⁽Reference Shrimali, Irons and Carlson²¹⁾. The amount of ³H-thymidine incorporated into both cell lines was measured after 60 h as an indication of the level of proliferation. Control cells incubated with both antibodies manifested a far more pronounced level of proliferation compared to unstimulated cells, whereas the level of proliferation of ΔTrsp^T T-cells was dramatically reduced under the same conditions (Fig. 3(A)). The data suggested that selenoprotein expression appeared to be essential for the ability of T-cells to proliferate in response to TCR stimulation.

Fig. 3. Activation of control and selenoprotein-deficient-cells (ΔTrsp^T) by T-cell receptor stimulation. T-cells were isolated from lymph nodes of control and ΔTrsp^T mice. (A) The level of proliferation was assessed by the incorporation of [³H]-thymidine in the absence or presence of anti-CD3/CD28 or anti-CD3 alone. The results of the averages of triplicate determinations from six control and ΔTrsp^T mice, respectively, are shown. (B and C) The levels of IL-2 production (B) and cell surface expression of IL-2 receptor α (C) in CD3/CD28-stimulated T-cells are shown. (D) Extracellular signal-regulated kinase (ERK) activation in T-cells stimulated with anti-CD3 and anti-CD28 was analysed at the time intervals indicated. Both phosphorylated (pERK) and total (ERK) proteins are shown. (E) Serum levels of major Ig at indicated time points were determined by ELISA in control and ΔTrsp^T mice following immunization with (4-hydroxy-3-nitrophenyl)acetyl-ovalbumin). Circles represent Ig levels of each animal. Bars represent the mean values. Experimental details are given in Shrimali et al.⁽Reference Shrimali, Irons and Carlson²¹⁾.

Since TCR activation is known to produce IL-2 in T-cells and generate T-cell IL-2 receptor expression, we initially examined IL-2 production in control and ΔTrsp^T T-cells (Fig. 3(B))⁽Reference Shrimali, Irons and Carlson²¹⁾. Both cell lines produced similar amounts of IL-2 in response to TCR activation. However, ΔTrsp^T T-cells manifested a much reduced IL-2 receptor induction compared to control cells (Fig. 3(C)).

Since TCR is coupled to multiple signalling pathways and it is known that mitogen-activated protein kinase extracellular signal-regulated kinase (ERK) links TCR activation to T-cell proliferation⁽Reference Pages, Guerin and Grall³⁰⁾, we examined ERK activation in control and selenoprotein-deficient T-cells in response to stimulation with CD3/CD28 antibodies⁽Reference Shrimali, Irons and Carlson²¹⁾. As shown in Fig. 3(D), T-cells from ΔTrsp^T mice manifested a very slight appearance of the phosphorylated or active forms of ERK following the treatment with anti-CD3 and anti-CD28 antibodies, while T-cells from control mice manifested a rapid appearance of the active forms of ERK.

The above data demonstrated that the loss of selenoprotein expression in T-cells resulted in defects in the development of functionally mature T-cells and their TCR-dependent activation. It is highly likely that such anomalies result in an impairment in T-cell-mediated immune responses. To further examine this possibility, control and ΔTrsp^T mice were immunized with NP-OVA ((4-hydroxy-3-nitrophenyl)acetyl-ovalbumin), an antigen that stimulates T-cell antibody production. Serum was taken from the immunized mice at different time periods after immunization and the levels of antigen-specific Ig were assessed by ELISA. Of the several Ig classes examined, IgM, IgG1, IgG2a, IgG2b and IgG3, the serum levels of antigen-specific antibodies were poorly raised in ΔTrsp^T mice (Fig. 3(E)). Thus, these data suggest that selenoproteins play an important role in T-cell immunity.

Reactive oxygen species in ΔTrsp^T T-cells

Since several selenoproteins are known to act as antioxidants and therefore have roles in protecting cells against oxidative stress, we examined whether selenoprotein-deficient T-cells produce higher levels of reactive oxygen species (ROS) than the corresponding control cells and how this may relate to T-cell defects in ΔTrsp^T mice. To elucidate the basal and TCR-stimulated ROS generation in ΔTrsp^T and control cells, we labelled the cells with cell-permeable, oxidation-sensitive DCFDA (2′-7′-dichlorofluorescin diacetate) dye. Interestingly, as shown in Fig. 4(A), the oxidation of DCFDA due to the generation of ROS was present to a greater extent in T-cells of ΔTrsp^T mice than in control mice even without TCR stimulation. TCR stimulation enhanced the ROS production in control cells, but had little effect on the ROS levels in ΔTrsp^T T-cells. The data suggest that selenoproteins have a major role in suppressing ROS production in T-cells and that the loss of this protein class results in T-cells being defective in TCR-induced and ROS-sensitive responses.

Fig. 4. Reactive oxygen species (ROS) production in T-cell-receptor-stimulated T-cells. (A) Control and ΔTrsp^T T-cells isolated and purified from lymph nodes and either unstimulated or stimulated with anti-CD3 antibody were used. ROS production was analysed by flow cytometry using 2′-7′-dichlorofluorescin diacetate. (B) Control and ΔTrsp^T T-cells were either unstimulated (None) or stimulated as indicated in the presence of various concentrations of N-acetyl cysteine (NAC). T-cell proliferation and all other determinations used in the figure are described in Shrimali et al.⁽Reference Shrimali, Irons and Carlson²¹⁾.

Finally, ΔTrsp^T T-cells were incubated in the presence of N-acetyl cysteine (NAC), which is known to block endogenously produced ROS, to assess whether this reagent can restore the ability of these cells to proliferate in response to TCR stimulation (Fig. 4B). The ability of the selenoprotein-deficient T-cells to proliferate in response to TCR stimulation was restored by NAC treatment. NAC alone did not induce T-cell proliferation. These observations suggest that selenoproteins are essential for reducing ROS generation in T-cells and that ROS production prevents T-cell activation.

Selenoprotein mRNA, selenoprotein expression and reactive oxygen species levels

To determine the relative expression levels of selenoproteins mRNA in mouse macrophages, the total RNA was isolated from bone marrow-derived normal macrophages and the expression of mRNA levels transcribed from all 24 mouse selenoprotein genes were analysed by real-time PCR. As shown in Fig. 5(A), relatively high levels of selenoprotein mRNA corresponding to GPx1, GPx4, TR1, Sep15, and selenoproteins P, R, K and T and selenophosphate synthetase 2 were expressed. Interestingly, the most highly expressed selenoprotein mRNA in macrophage, like that in T-cells (see earlier), was GPx1 and a major role of this selenoenzyme is as an antioxidant. Similarly, Sep15, GPx4 and selenoproteins P, K and T were also highly expressed in macrophages.

Fig. 5. Selenoprotein gene expression in macrophages. (A) Selenoprotein gene expression in macrophages was analysed by real-time PCR and is shown as the relative mRNA level to that of Gusb which was used as the internal control. (B) Northern blot analysis showing levels of Sec tRNA in control and ΔTrsp macrophage. (C) ⁷⁵Se-labelled selenoproteins were visualized by autoradiography after SDS electrophoresis in control and ΔTrsp macrophage as shown in the right panel. The left panel shows Coomassie Blue-stained (CBB) gel, which served as a loading control. The identities of the major labelled selenoproteins are designated on the right of the panel. (D) Macrophages were stained with 2′-7′-dichlorofluorescin diacetate (DCFDA) and reactive oxygen species (ROS) production analysed by confocal microscopy. Fluorescence (left panels) and phase contrast (right panels) images are shown. (E) Macrophages were stained with DCFDA and ROS production analysed by flow cytometry. Experimental details of the studies shown in this figure are given in Carlson et al.⁽Reference Carlson, Yoo and Sano²²⁾.

We deleted Trsp in macrophage (ΔTrsp^M) and examined the effects of Trsp loss on selenoprotein expression and function. Initially, we demonstrated that Trsp was no longer expressed in this cell type (Fig. 5(B)). Selenoprotein expression was also examined in macrophages by incubating the bone marrow-derived macrophages from control and ΔTrsp^M mice in the presence of ⁷⁵Se and electrophoresing a protein extract on a polyacrylamide gel (Fig. 5(C), left panel). Several selenoproteins whose mRNA were also detected by real-time PCR were visualized on the gel, while the expression of selenoproteins in ΔTrsp^M cells were almost completely abolished (Fig. 5(C), right panel). The low levels of Sec tRNA and selenoproteins observed in ΔTrsp^M macrophages in Fig. 5(B and C) are likely due to the contamination of the bone marrow-derived macrophage preparations with non-macrophage cells as has been observed in studies involving other cell types⁽Reference Carlson, Novoselov and Kumaraswamy²⁷⁾.

The functions of only about half of selenoproteins have been characterized and most are oxidoreductases and/or antioxidants⁽Reference Hatfield, Carlson and Xu³¹⁾. Since selenoprotein-less T-cells accumulate higher ROS levels than the corresponding control cells (see Fig. 4(A)), ROS levels in ΔTrsp^M macrophages were also assessed by staining (Fig. 5(D and E)). Resting selenoprotein-deficient macrophages manifested higher steady state ROS levels than control macrophages.

In vitro and in vivo inflammatory responses

Due to the highly significant role that macrophages have in initiating inflammation, it was important to assess the ability of the corresponding selenoprotein-deficient T-cells in the inflammation process. Macrophages generate cytokines in response to tissue injury and microbial infection. The fact that several inflammatory stimuli such as lipopolysaccharide (LPS) cause transient ROS accumulation as part of cellular signalling events⁽Reference Hsu and Wen^32, Reference Park, Jung and Park³³⁾, and several selenoproteins, e.g. GPx isoforms, which are known to regulate inflammatory responses⁽Reference Brigelius-Flohe, Friedrichs and Maurer^34, Reference Esworthy, Aranda and Martin³⁵⁾, prompted us to examine the effect of selenoprotein deficiency and the resulting redox imbalance on macrophage inflammatory response. We initially treated control and ΔTrsp^M macrophages with LPS and examined the degradation and replenishment of IκBα. It should be noted that the transcription factor NF-κB, and mitogen-activated protein kinase (ERK, c-Jun N-terminal kinase and p38) signalling pathways are essential for cellular responses to inflammatory stimuli and that both the degradation and replenishment of IκBα are indicative of NF-κB activation. However, in LPS-treated control and ΔTrsp^M macrophages, the degradation and replenishment of IκBα occurred independently of the selenoprotein status and the induction of the active forms of the protein kinases (i.e., phosphorylated ERK, JNK and p38) was similar and therefore considered normal⁽Reference Carlson, Yoo and Sano²²⁾. Furthermore, comparison of LPS-induced inflammatory gene expression in the two bone marrow-derived macrophage groups and analyses of gene expression by real-time PCR revealed that the magnitude and kinetics of expression of the genes encoding the chemokines KC (Cxcl1), macrophage inflammatory protein-2 (Cxcl2) and the cytokines TNF-α and IL-1β were similar⁽Reference Carlson, Yoo and Sano²²⁾.

We next turned our attention to examining the effects of selenoprotein deficiency in macrophages on the pathology of mice by exposing control and ΔTrsp^M mice to various models of inflammatory response. The models used included zymosan-induced peritonitis, LPS endotoxemia and chemical irritant (12-O-tetradecanoylphorbol-13-acetate) dermatitis. In these models, control and ΔTrsp^M mice had comparable rates of neutrophil infiltration, rates of mortality and levels of cytokine production and local edema formation. Since in vitro and in vivo inflammatory responses in both control and ΔTrsp^M macrophages were similar, we have not shown the data, but refer the reader to the original work⁽Reference Carlson, Yoo and Sano²²⁾. Interestingly, the loss of selenoprotein expression and the deregulated ROS generation in ΔTrsp^M macrophages did not appear to cause any detectable detrimental effects concerning inflammatory responses.

Macrophage microarray analysis

Since there were unexpectedly little or no distinguishable inflammatory phenotypes in ΔTrsp^M mice, it seemed reasonable that selenoproteins may have other functions in macrophages in addition to roles in inflammatory gene expression. To explore this possibility, an analysis of the global gene expression profiles was carried out in control and ΔTrsp^M macrophages. We used whole-genome DNA microarrays from mice to assess the expression levels of mRNA in the two macrophage groups. Those genes that manifested a significantly higher or lower expression in ΔTrsp^M macrophages compared to control macrophages are shown in Fig. 6(A) and are designated by those dots (designating mRNA) that fell outside the lines above and below normal expression levels indicating significant changes.

Fig. 6. Microarray analysis of ΔTrsp^M macrophages. (A) Genes that were up-regulated and down-regulated in ΔTrsp^M macrophages are shown in the scatter plot. The ordinate and abscissa represent the logarithmic values of the signal intensity of individual microarray spots. The upper and lower cut-off lines illustrate the margins of gene expression ratio wherein 3-fold or higher up-regulated genes (spots above the light grey line) and 3-fold or lower down-regulated genes (spots below the dark black line) are shown as outside the upper and lower cut-off lines, respectively. (B) Those genes whose expression levels were identified in DNA microarray analysis manifesting an alteration greater than 3-fold were validated by real-time PCR. The relative expression of genes in macrophages that were either untreated or treated with lipopolysaccharide is shown. (C) Expression of genes and protein products that were up-regulated by selenoproteins is shown. Experimental details of the studies shown in this figure are given in Carlson et al.⁽Reference Carlson, Yoo and Sano²²⁾.

The expression levels of those mRNA that showed significant changes in selenoprotein-deficient macrophages were verified by real-time PCR analysis (Fig. 6(B)). Interestingly, many genes that deviated significantly in their expression levels from normal in ΔTrsp^M macrophages were functionally related to the formation, remodelling cellular interaction with the extracellular matrix (ECM). Each of the ECM-related genes that we found to be significantly altered in ΔTrsp^M macrophages were up-regulated compared to the corresponding control cells (Fig. 6(C)). These included genes that encode ECM components (collagen chains, extracellular proteoglycans and secreted glycoproteins; Col1a1, Col5a2, Bgn and Ctgf), inhibitors of ECM proteolysis (metalloendopeptidase and serine-type endopeptidase inhibitors; Timp3 and Serpinh1) and ECM-induced cytoskeletal remodelling (actin-binding proteins; Tagln, Enah and Cald1).

Selenoprotein deficiency and macrophage invasiveness

The microarray analysis of ΔTrsp^M macrophages suggested that macrophage–ECM interactions were altered by the lack of selenoprotein expression in such a manner that matrix remodelling was decreased and the surrounding ECM was reinforced. Such effects might be expected to impair the migration of macrophages through the ECM and basement membrane. To test this possibility, we compared the invasiveness of control and ΔTrsp^M macrophages in a protein gel matrix. The migration of selenoprotein-deficient macrophages was dramatically reduced in the gel-laden transwell chambers (Fig. 7(A)). We repeated this assay, but without the gel in the transwell chamber to assess whether the gel invasion phenotype arose from a cell-intrinsic motility defect. Surprisingly, the migration of ΔTrsp^M macrophages was slightly higher than control macrophages (Fig. 7(B)) suggesting that the changes observed in ECM-related gene expression in macrophages appeared to decrease the ability of the cell to migrate only in surroundings consisting of ECM components.

Fig. 7. Analysis of ΔTrsp^M macrophage invasion in a protein gel matrix. (A) Macrophage migration was examined by a transwell assay with a protein gel matrix (Matrigel) layer in the upper chamber. Numbers below each panel indicate the relative migration of macrophages and they represent the mean and standard deviation (n 3). Numbers in parenthesis indicate the percentage of cell migration in the ΔTrsp^M sample relative to the control sample. (B) Macrophage migration was examined by a transwell assay in the absence of a Matrigel layer and the data are presented as in (A). Experimental details of the studies shown in this figure are given in Carlson et al.⁽Reference Carlson, Yoo and Sano²²⁾.