Transferrin is required for early T-cell differentiation

Abstract

Transferrin, the major plasma iron carrier, mediates iron entry into cells through interaction with its receptor. Several in vitro studies have demonstrated that transferrin plays an essential role in lymphocyte division, a role attributed to its iron transport function. In the present study we used hypotransferrinaemic (Trf^hpx/hpx) mice to investigate the possible involvement of transferrin in T lymphocyte differentiation in vivo. The absolute number of thymocytes was substantially reduced in Trf^hpx/hpx mice, a result that could not be attributed to increased apoptosis. Moreover, the proportions of the four major thymic subpopulations were maintained and the percentage of dividing cells was not reduced. A leaky block in the differentiation of CD4⁻ CD8⁻ CD3⁻ CD44⁻ CD25⁺ (TN3) into CD4⁻ CD8⁻ CD3⁻ CD44⁻ CD25⁻ (TN4) cells was observed. In addition, a similar impairment of early thymocyte differentiation was observed in mice with reduced levels of transferrin receptor. The present study demonstrates, for the first time, that transferrin itself or a pathway triggered by the interaction of transferrin with its receptor is essential for normal early T-cell differentiation in vivo.

Introduction

Transferrin (Trf), the major plasma iron carrier, is synthesized and secreted mainly by the liver. Trf delivers iron to tissues through binding to its receptor (TrfR) and receptor-mediated endocytosis (reviewed in ¹). It is widely believed that iron delivery is the primary function of Trf. Strongly conserved ontogenetic and phylogenic patterns of expression suggest that the Trf/TrfR-mediated iron delivery pathway is necessary for proper development. This was confirmed by the observation that TrfR-null mice die in utero.² Heterozygous mice lacking one TrfR allele (TrfR^+/–) are generally similar to their wild-type littermates; they are not anaemic, but have microcytic erythrocytes and decreased liver iron stores.

Transferrin was identified more than two decades ago as a critical growth factor for peripheral lymphocytes in vitro.^3,4 Additional studies demonstrated that proliferating thymocytes and lymphocytes express high levels of TrfR^5,6 and anti-TrfR antibodies inhibit lymphocyte and thymocyte proliferation and differentiation in vitro.^6,7

Hypotransferrinaemic (Trf^hpx/hpx) mice carry a spontaneous point mutation in a splice donor site of the Trf gene, which results in severe deficiency in serum transferrin (less than 1% of the normal levels).⁸ These mice are not viable unless Trf is exogenously administered during the first 4 weeks after birth.⁹ While heterozygous hypotransferrinemic (Trf^hpx/+) mice have half normal levels of circulating Trf and do not differ greatly from wild-type animals, homozygous mice have evident abnormalities of iron metabolism from birth. Specifically, homozygous mice are profoundly anaemic, because of impaired assimilation of iron by erythroid precursors, and have iron overload in parenchymal tissues as a result of increased iron absorption and increased clearance of non-transferrin-bound plasma iron. Heterozygosity for one wild-type allele prevents anaemia and delays iron accumulation in tissues until the sixth month of age.⁹

We have taken advantage of well-characterized hypotransferrinaemic and TrfR^+/– mice to study the role of Trf in T-cell differentiation in vivo.

Materials and methods

Animal care and genotyping

Animals were maintained under a 12-hr light/dark cycle and fed standard rodent chow and tap water ad libitum. They were handled in accordance with the European Union rules and the National Institute of Health guidelines for the care and handling of laboratory animals.

Trf^hpx/hpx mice and littermate controls, heterozygous (Trf^hpx/+) and wild-type (Trf^+/+) mice, were obtained from crosses of heterozygous BALB/cJ-Trf^hpx/+ mice generously provided by Dr Jerry Kaplan (University of Utah). Trf^hpx/hpx mice were easily identified from birth because of their anaemic pallor. To distinguish between heterozygous and wild-type mice, all animals were genotyped by polymerase chain reaction (PCR) using forward and reverse primers: 5′-TCAAGTCCACCACCAAGGACC-3′ and 5′-CTGGGTTTCTCTGCGACCTCTC-3′, respectively. PCR products were digested with the restriction enzyme Hpy CH4 IV (New England BioLabs, Beverly, MA), which recognizes a site present in wild-type alleles and absent in mutant alleles.

Trf^hpx/hpx animals received weekly intraperitoneal injections of 6 mg human Trf (Roche, Indianapolis, IN) in Iscove's modified Dulbecco's medium (IMDM), from the first day of life until the 4th week of age. Wild-type and heterozygous mice were given the same treatment with IMDM lacking Trf (Gibco, Paisley, UK). Mice were analysed 8–13 weeks after the last transferrin treatment.

TrfR^+/– mice, on a 129SvEvTac genetic background, were 16 weeks old and genotyped as described previously.²

Iron, transferrin and haemoglobin measurements

Blood was collected into heparinized capillary tubes by orbital puncture. Haemoglobin was determined in a blood cell counter (Sysmex, Japan). Serum iron-transferrin concentration was determined by the Ferrozine method and human transferrin was evaluated by an immunoturbidimetric method, using the COBAS-Integra 800 autoanalyser (Roche, Switzerland). The relative amount of mouse transferrin was determined by Western blot analysis using rabbit anti-mouse transferrin (Research Diagnostics Inc, Flanders, NJ), and mouse Trf as control (Sigma, St. Louis, MO).

Antibodies and flow cytometry analysis

Single lymphoid cell suspensions were prepared and cells were labelled with the following monoclonal antibodies: fluoroscein isothiocyanate (FITC)-labelled anti-CD8α, peridinin chlorophyll-a protein (PerCP)-labelled anti-CD8α, phycoerythrin (PE)-labelled anti-CD4, allophycocyanin (APC)-labelled anti-CD3ɛ, PE-labelled anti-CD44, APC-labelled anti-CD25 (PharMingen, San Diego, CA), and FITC-labelled anti-CD24, APC-labelled anti-CD25, biotin-labelled anti-CD69 (Caltag, Burlingame, CA). To analyse the expression of CD25 and CD44 on CD4^– CD8^– CD3^– (TN) thymocytes, cells were labelled with biotinylated antibodies: anti-CD3ɛ, anti-CD4, anti-CD8α (PharMingen), anti-T-cell receptor (TCR)αβ (Caltag) and identified with PE-Cy5 coupled to streptavidin (DAKO, Glostrup, Denmark). Dead cells, identified by their light-scattering characteristics, were excluded from analysis. All data acquisition and analysis was performed with a FACSCalibur^™ (Becton Dickinson, San Jose, CA) instrument interfaced with Macintosh CELLQuest software.

Bromodeoxyuridine incorporation and detection

Twenty and 16 hr before death, the mice received intraperitoneal injections of 1 mg bromodeoxyuridine (BrdU) (Sigma). Thymocytes identified by surface staining were fixed and permeabilized in 1% paraformaldehyde, 0·01% Tween-20 in phosphate-buffered saline (PBS) for 40 h at 4°, and then subjected to a BrdU DNase treatment detection technique previously described¹⁰ using FITC-conjugated anti-BrdU antibody (Becton Dickinson).

Thymocyte apoptosis detection

Single cell suspensions were stained with APC-labelled anti-CD8α and PE-labelled anti-CD4 antibodies (PharMingen). To detect apoptotic thymocytes, cells were then labelled for annexin V, using the annexin V–FITC apoptosis detection kit (PharMingen), according to the manufacturer's instructions.

For detection of DNA fragmentation in situ, paraffin-embedded sections were analysed by the TdT-mediated biotin–dUTP nick-end labelling (TUNEL) method using the ApoTag Plus Peroxidase in situ Apoptosis detection kit (Chemicon, Norcross, GA) according to the manufacturer's protocol.

Statistical analysis

The statistical analysis focused on comparisons between groups of mice with different genotypes. Student's t-test with P < 0·05 was used to define statistically significant differences between groups

Results

Iron, transferrin and haemoglobin levels in hypotransferrinaemic mice

Trf^hpx/hpx mice are anaemic because of impaired delivery of iron to red cell precursors. The haemoglobin levels of the Trf^hpx/hpx mice used in this study confirmed the anaemic status of these mice (Trf^hpx/hpx: 3·4 ± 1·2; Trf^hpx/+: 16·4 ± 0·6 and Trf^+/+: 16·2 ± 0·7 g/dl). Serum iron levels were also significantly reduced in the transferrin deficient mice (Trf^hpx/hpx: 75 ± 59; Trf^hpx/+: 237 ± 30 and Trf^+/+: 309 ± 73 μg/dl). Trf^hpx/hpx mice were treated during the first 4 weeks of life with human transferrin and studied 8–13 weeks after the last transferrin injection. As expected, serum human transferrin levels were under the detection limit at the time of study (data not shown). Western blot analysis confirmed previous results showing that the expression of Trf in the Trf^hpx/hpx is lower than 1% of the levels observed in wild-type animals (data not shown).⁸

Hypotransferrinaemic mice display reduced absolute thymus cell numbers but no alterations in the proportion of the four main thymic populations

Analysis of thymus absolute cell number revealed that a 50% decrease in plasma Trf, observed in Trf^hpx/+ mice, does not affect the number of thymocytes produced, but near total Trf deficiency, present in the Trf^hpx/hpx mice, leads to major differences in thymocyte numbers. Figure 1(a) presents data from 16-week-old mice (Trf^hpx/hpx: 29 ± 5 × 10⁶; Trf^hpx/+: 73 ± 15 × 10⁶ and Trf^/: 67 ± 17 × 10⁶). Similar results were observed at 12 weeks.

Figure 1.

Reduced thymic cellularity in Tfr^hpx/hpx mice with no major alterations in the four main thymic subpopulations.(a)Each dot represents the absolute thymocyte number from one mouse analysed. Data are shown for 16-week-old mice. Similar results were observed for 12-week-old mice.(b)Histograms show the percentage of TN, DP, CD8SP and CD4SP thymocytes for each genotype analysed. Genotypes differences are specified by black and white shading. Results show the mean ± SD values obtained from at least eight mice in each group. Data presented correspond to 16-week-old mice, similar results were observed for 12-week-old mice.

Despite reduced cell numbers, the proportions of the four major thymic populations (CD4^– CD8^– CD3^– (TN), CD4⁺ CD8⁺ (DP), CD4⁺ CD8^– (CD4SP) and CD4^– CD8⁺ (CD8SP)) were not affected by reduced Trf levels (Fig. 1b). A slight but not statistically significant increase in the percentage of TN cells was consistently observed in Trf^hpx/hpx mice when compared to Trf^hpx/+ or wild-type mice (Trf^hpx/hpx: 2·3 ± 0·4; Trf^hpx/+: 1·8 ± 0·3; Trf^+/+: 1·9 ± 0·4). Thymocytes from the four major thymus populations were further characterized for the expression of additional T-cell differentiation markers. No differences were observed on the expression patterns of CD69, CD24 and CD25 (data not shown). Peripheral CD4⁺ and CD8⁺ T-cell numbers are not affected by severe deficiency of serum transferrin (Table 1).

Table 1.

Number of T cells (× 10⁶) in inguinal lymph nodes from Trf^hpx/hpx mice

Genotype	CD4	CD8	CD4/CD8
Trf +/+ (n = 14)	1·4 ± 0·9	0·5 ± 0·2	3·0 ± 0·9
Trf hpx/hpx (n = 22)	1·0 ± 0·7	0·4 ± 0·2	2·6 ± 0·6

Data represent means and their standard deviations.

Reduced thymus cellularity in hypotransferrinaemic mice is not caused by altered proliferation or increased apoptosis

To investigate whether the reduced absolute cell number found in Trf^hpx/hpx thymus (Fig. 1a) resulted from reduced thymocyte proliferation, thymocyte BrdU incorporation was measured in Trf^hpx/hpx, Trf^hpx/+ and wild-type mice. BrdU is a thymidine analogue that is incorporated into the DNA of dividing cells. As no differences were found in the number and proportion of thymocyte populations between Trf^hpx/+ and wild-type animals, we combined the results from these animals and referred to them as controls. No reduction was observed in the proliferation rate of Trf^hpx/hpx thymocytes when compared to the controls (Fig. 2). Figure 2(a) shows a representative example of BrdU incorporation in thymocytes from one wild-type and one Trf^hpx/hpx animal. Surprisingly, as shown in Fig. 2(b), we found that the percentage of dividing thymocytes in the TN population was significantly increased in Trf^hpx/hpx mice when compared to control mice (Trf^hpx/hpx: 37·1 ± 3·3; control: 27·1 ± 5·1; P < 0·01).

Figure 2.

Thymocyte proliferation and apoptotic levels in Trf^hpx/hpx mice. (a)Comparison of TN, DP, CD4SP and CD8SP thymocyte proliferation measured by BrdU incorporation, between Trf^hpx/hpx and wild-type 17-week-old mice. Numbers in each panel indicate percentage of BrdU⁺ cells. (b) Histograms show the percentage of BrdU⁺ TN, DP, CD4SP and CD8SP thymocytes for Trf^hpx/hpx and control mice (Trf^hpx/+ and wild-type mice were used as controls). Genotypes differences are indicated by black and white colouring. Results show the mean ± SD values obtained from five mice in each group. Data presented correspond to 16–17-week-old mice. The significance of the differences between groups was tested by Student's t-test (*P = 0·03). (c)Annexin V staining of, CD4, CD8 and CD3 labelled, thymocytes from 16-week-old Trf^hpx/hpx mice (black line n = 3) and control mice (grey line n = 4).

To directly assess thymocytes for possible defects in survival, we examined the spontaneous rates of apoptosis in thymuses and freshly isolated thymocytes from Trf^hpx/hpx mice and control mice (Trf^hpx/+ and Trf^+/+ mice). TUNEL analysis demonstrated no differences in the level of apoptosis in the 16-week-old mice (data not shown). Annexin V staining of total thymocytes confirmed that cells from Trf^hpx/hpx thymus do not have an elevated apoptotic index. This was true for each of the four main thymocyte subsets (Fig. 2c).

Low levels of transferrin are associated with a leaky block at early thymocyte differentiation

To further characterize thymic differentiation, we studied the phenotype of the most immature thymocytes. The differentiation of CD4⁻ CD8⁻ thymocytes is marked by the transient expression of CD44 and CD25. TN thymocytes advance progressively through CD44⁺ CD25^– (TN1), CD44⁺ CD25⁺ (TN2) and CD44^– CD25⁺ (TN3) stages to finally give rise to CD44^– CD25^– cells (TN4), which are the immediate precursors of the DP thymocytes.¹¹ As shown in Fig. 3(a), Trf^hpx/hpx mice had a smaller percentage of cells reaching the TN4 differentiation stage. Consequently, there was a severe reduction in the absolute number of TN4 thymocytes (Fig. 3b). No differences between groups were seen in the absolute number of the most immature thymocytes (TN1 thymocytes).

Figure 3.

Trf^hpx/hpx mice show a leaky block in the transition between the TN3 to TN4 stage of thymocyte differentiation. (a) TN thymocytes were analysed for the expression of CD44 and CD25. A representative example of a 12-week-old mice is shown. Similar results were observed for 16-week-old mice. Numbers in each panel indicate percentage of cells. (b) Similar numbers of TN1 thymocytes give rise to clearly less TN4 cells in the Trf^hpx/hpx mice. Histograms show absolute numbers of TN1 and TN4 thymocytes. Genotypes differences are indicated by black and white shading. Results show mean ± SD values of: Trf^+/+n = 3, Trf^hpx/+n = 7, Trf^hpx/hpxn = 6. Data presented correspond to 16-week-old mice, similar results were observed for 12-week-old mice. The significance of the differences between wild-type and Trf^hpx/hpx mice was tested by Student's t-test (*P < 0·002).

Reduced expression of transferrin receptor impairs early thymocyte differentiation

To further understand the role of Trf in thymocyte differentiation, we studied thymuses from mice heterozygous for a TrfR-null mutation (TrfR^+/–). TrfR^–/– mice are not viable.² It was previously described that 11-week-old TrfR^+/– mice are not anaemic although erythrocytes are microcytic.² The same holds true for older mice used in this study: haemoglobin levels of TrfR^+/– mice (14·7 ± 0·6 μg/dl) were not significantly different from control mice (15·2 ± 0·7 μg/dl), and microcytosis persisted.

Although not reaching statistical significance, the TrfR^+/– mice had reduced number of thymocytes TrfR^+/– (145 ± 34 × 10⁶, n = 5) compared to wild-type littermates (187 ± 40 × 10⁶, n = 5). In addition, thymuses of these mice had a lower proportion of TN thymocytes and a slight but consistent increased proportion of the DP population (Fig. 4a). Thymocytes, from the four major thymus populations, were further characterized for the expression of additional T-cell differentiation markers. No differences were observed on the expression patterns of CD69, CD24 and CD25 (data not shown).

Figure 4.

TrfR^+/– mice show a leaky block in the transition between the TN3 to TN4 stage of thymocyte differentiation. (a)Histograms show the percentage of TN, DP, CD8SP and CD4SP thymocytes for TrfR^+/– (n = 5) and control mice (n = 4). Genotypes differences are indicated by black and white colouring. Results show the mean ± SD values. (b)TN thymocytes were analysed for the expression of CD44 and CD25. A representative example of three independent experiments is shown. Numbers in each panel indicate percentage of cells. (c)Histograms show absolute numbers of TN1 and TN4 thymocytes for TrfR^+/– (n = 4) and control mice (n = 3). Genotypes are specified by colour coding. Results show mean ± SD values. Data presented correspond to 16-week old-mice. The significance of the differences between groups was tested by Student's t-test (*P < 0·03; **P < 0·002).

Analysis of the most immature thymocytes revealed that TrfR^+/– mice have the same leaky block in early thymocyte differentiation as the Trf^hpx/hpx mice: they have a smaller percentage of cells reaching the TN4 differentiation stage (Fig. 4b) with a consequent reduction in the absolute number of TN4 thymocytes (Fig. 4c). In contrast to Trf^hpx/hpx mice, lower numbers of TN1 thymocytes were present in the TrfR^+/– mice. No difference in the number or ratios of CD4⁺ and CD8⁺ T cells was seen in the periphery.

Discussion

We have used two different mouse models to investigate whether transferrin, alone or through interaction with its receptor, plays an important role in normal T-cell differentiation in vivo. Profound deficiency of circulating Trf or decreased expression of TrfR result in a severe reduction in the number of thymocytes that differentiates from the stage TN3 into TN4. We also observed a reduction of the total number of thymocytes, but the two animal models differed: a strong reduction was observed in the Trf^hpx/hpx mice while TrfR^+/– mice have a minor reduction. Furthermore, we demonstrated that this reduction does not result from the inability of thymocytes to divide normally or from increased apoptosis. BrdU incorporation was not reduced, and no differences were found on markers of apoptosis in situ and ex vivo. Since the absolute number of TN1 thymocytes was similar in Trf^hpx/hpx, Trf^hpx/+ and wild-type mice, the strong reduction in absolute thymocyte numbers does not seem to be a consequence of altered migration of thymocyte precursors from the bone marrow. The observation that TrfR^+/– mice had a decreased number of TN1 thymocytes indicate that an effect of the transferrin receptor in bone marrow T-cell differentiation could not be excluded.

Interestingly, no changes in thymocyte phenotype were observed in Trf^hpx/+ mice, suggesting that half normal Trf plasma levels are adequate for normal thymic differentiation. Reduction of the number of transferrin receptors, on the other hand, appeared to be sufficient to impair thymocyte differentiation. Based on these observations, we speculate that a normal complement of transferrin receptors is required for thymocyte differentiation, but that transferrin is normally present in excess.

The influence of transferrin and/or its receptor seems to be specific to thymocyte early differentiation, as we found no differences in the number of CD4⁺ and CD8⁺ peripheral lymphocytes in both Trf^hpx/hpx and TrfR^+/– mice.

Iron is known to be essential for cell division. In the present study, there was no reduction in thymocyte proliferation in Trf^hpx/hpx mice, suggesting that iron delivery to thymocytes is adequate in these animals. Since it seems unlikely that 1% normal Trf levels can provide sufficient iron to thymocytes through TrfR-mediated endocytosis, we suggest that the thymus, like other organs in these mice, uses non-Trf bound iron uptake mechanisms to fulfill its iron needs.¹

As a consequence of hypotransferrinaemia, Trf^hpx/hpx mice develop tissue iron overload and anaemia because of impaired iron uptake by erythroid precursors. Anaemia and iron overload cannot explain however, the leaky block observed during early thymic differentiation. The TrfR^+/– mice show the same type of block but do not present anaemia or iron overload.²

A recent paper by Ned et al. reports the involvement of TrfR in lymphopoiesis.¹² Ned's observations provide additional support for the present conclusions.

Consistent with our results, a previous study showed that treatment of fetal thymic organ cultures with an anti-TrfR antibody led to a nearly absolute block of differentiation from CD4^– CD8^– (DN) to DP cells.⁷ In those experiments, the numbers of DN CD25⁺ cells (roughly corresponding to TN2 and TN3 cells in our study) were not diminished by antibody treatment, suggesting that TrfR has a role in the terminal differentiation of DN thymocytes.⁷

Thymocyte maturation arrest between TN3 and TN4 has previously been observed in several animal models. Mice lacking expression of proteins involved in surface expression of pre-TCR or in T-cell signalling pathways (including RAG,^13,14 CD3ɛ,^15,16 Lck/Fyn,^17,18 ZAP-70/Syk,¹⁹ SLP-76^20,21 and LAT²²) have altered thymus differentiation (reviewed in ²³). In all of these mouse models, thymocytes fail to develop beyond the TN3 stage. It is believed that blockage at this step in differentiation is a result of a requirement for expression of a pre-TCR complex which is coupled to an intracellular signalling pathway. It is interesting to note that TrfR has been implicated in T-cell activation^24–26 specifically through its association with the TCRζ chain, which interacts with ZAP70 tyrosine kinase upon ligand binding.²⁶ It is also worth noting that TrfR stimulation by an anti-TrfR antibody or by iron-loaded Trf leads to tyrosine phosphorylation events similar to those triggered by TCR stimulation.²⁶ It is therefore possible that the phenotype we observe in the Trf^hpx/hpx mice results from altered signal transduction. If so, severely reduced levels of circulating Trf (<1%) are insufficient to fully support normal Trf mediated pre-TCR signalling pathways. It is important to note that pre-TCR signalling enhances proliferation.²⁷ TN cells from Trf^hpx/hpx mice divide more compared to wild-type TN thymocytes. If the increased proliferation occurs after pre-TCR expression it would argue for a positive, and not negative, role of pre-TCR signalling in T-cell differentiation in Trf^hpx/hpx mice. Future experiments dissecting proliferation on the TN subpopulations should elucidate this point.

In conclusion, our studies of Trf^hpx/hpx and TrfR^+/– mice indicate that Trf, via its receptor, is required for normal early T-cell differentiation, specifically in the progression from TN3 to the TN4 thymocytes. We hypothesize that the role of Trf in thymus differentiation is mediated by pre-TCR-dependent signaling pathways.