Metabolic and immunologic consequences of limited adenosine deaminase expression in mice.

Adenosine deaminase (ADA; EC 3.5.4.4) deficiency in humans is an autosomal recessive genetic disorder that results in severe combined immunodeficiency disease. ADA-deficient mice generated by targeted gene disruption die perinatally, preventing postnatal analysis of ADA deficiency. We have recently rescued ADA-deficient fetuses from perinatal lethality by expression of an ADA minigene in the placentas of ADA-deficient fetuses, thus generating postnatal mice admissible to analysis of ADA deficiency. The minigene used also directed ADA expression to the forestomach postnatally, producing adult animals that lacked ADA enzymatic activity in all tissues outside the gastrointestinal tract. Mice with limited ADA expression exhibited profound disturbances in purine metabolism, including thymus-specific accumulations of deoxyadenosine and dATP, and inhibition of S-adenosylhomocysteine hydrolase in the thymus, spleen, and, to a lesser extent, the liver. Lymphopenia and mild immunodeficiency were associated with these tissue-specific metabolic disturbances. These mice represent the first genetic animal model for ADA deficiency and provide insight into the tissue-specific requirements of ADA.

Adenosine deaminase (ADA; EC 3.5.4.4) deficiency in humans is an autosomal recessive genetic disorder that results in severe combined immunodeficiency disease. ADA-deficient mice generated by targeted gene disruption die perinatally, preventing postnatal analysis of ADA deficiency. We have recently rescued ADA-deficient fetuses from perinatal lethality by expression of an ADA minigene in the placentas of ADA-deficient fetuses, thus generating postnatal mice admissible to analysis of ADA deficiency. The minigene used also directed ADA expression to the forestomach postnatally, producing adult animals that lacked ADA enzymatic activity in all tissues outside the gastrointestinal tract. Mice with limited ADA expression exhibited profound disturbances in purine metabolism, including thymus-specific accumulations of deoxyadenosine and dATP, and inhibition of S-adenosylhomocysteine hydrolase in the thymus, spleen, and, to a lesser extent, the liver. Lymphopenia and mild immunodeficiency were associated with these tissue-specific metabolic disturbances. These mice represent the first genetic animal model for ADA deficiency and provide insight into the tissue-specific requirements of ADA.
Genetic defects in purine metabolism in humans result in serious metabolic disorders (1), often with tissue-specific phenotypes. A particularly striking example of this is adenosine deaminase (ADA) 1 deficiency, which if untreated results in severe combined immunodeficiency disease (2). Although ADA deficiency was the first of the inherited immunodeficiencies for which the underlying genetic defect was identified (3), more than 25 years of subsequent research has not yielded a completely satisfactory explanation for the metabolic mechanisms which lead to the lymphoid specificity of the disease. However, available evidence suggests that the metabolic basis for ADAdeficient immunodeficiency is related to the physiological impact of the ADA substrates, adenosine and deoxyadenosine ( Fig. 1) (2,4). Adenosine functions as an extracellular signal transducer that mediates a variety of physiological effects by binding to adenosine receptors present on the surface of target cells (5). Deoxyadenosine behaves as a cytotoxic metabolite and is believed to provide the metabolic basis for ADA-deficient immunodeficiency (2). There are two mechanisms through which deoxyadenosine is thought to be lymphotoxic (Fig. 1). In one case, the accumulation of intracellular deoxyadenosine interferes with deoxynucleotide synthesis via its phosphorylation to dATP and subsequent inhibition of ribonucleotide reductase (6 -10). The other mechanism involves inhibition of the enzyme S-adenosylhomocysteine (AdoHcy) hydrolase by deoxyadenosine (11). Inhibition of AdoHcy hydrolase in turn leads to the accumulation of AdoHcy which acts as an inhibitor of S-adenosylmethionine (AdoMet)-mediated transmethylation reactions (12). Although these pathways appear to be accessed in ADAdeficient cells, the mechanisms involved in selective T and B cell elimination are not fully understood.
Biochemical hallmarks of ADA deficiency in humans are consistent with these proposed mechanisms of deoxyadenosine cytotoxicity. They include increases in plasma adenosine and deoxyadenosine (13), elevations of deoxyadenosine in the urine, marked accumulations of dATP in erythrocytes (10,14), and inhibition of erythrocyte AdoHcy hydrolase (12). A smaller population of ADA-deficient patients has been identified with late/delayed-onset of combined immunodeficiency disease (15). These patients harbor mutations that allow for low levels of ADA enzymatic activity, resulting in less severe accumulations of ADA substrates and the persistence of some T and B cells. This supports the hypothesis that the severity of the disease relates to the severity of substrate accumulations. However, exactly how the disturbances in deoxynucleotide metabolism and AdoMet-mediated transmethylation lead to the immunodeficiency seen in the absence of ADA is still unclear. Furthermore, it has been difficult to correlate metabolic disturbances with specific tissues due to their inaccessibility in humans. Efforts to understand the metabolic and tissue-specific basis for the immunodeficiency would likely benefit from the development of an ADA-deficient animal model in which the metabolic effects associated with ADA deficiency can be easily studied in the relevant tissues.
Attempts to generate ADA-deficient mice have recently been conducted by two groups, producing animals with independent sites of gene disruption (16,17). In both cases similar phenotypes were observed. Heterozygous matings produced Ada-null fetuses that died perinatally in association with severe liver damage, incomplete expansion of the lungs, and small intestine cell death (16,17). Liver damage was evident by 16.5 days post-coitum and worsened through 18.5 days post-coitum, preceding death of the fetuses (16). This phenotype was accompanied by pronounced disturbances in purine metabolism (16,17). The lymphoid organs of ADA-deficient fetuses and newborn pups were not largely affected, although there were minor reductions in the number of CD4-positive and CD8-positive lymphoid cells in livers of ADA-deficient fetuses (16). These observations suggest that ADA is essential for fetal survival in mice; however, they precluded our ability to assess the role of ADA in the postnatal immune system.
During fetal stages of development in the mouse, greater than 95% of ADA enzymatic activity at the gestation site is found in the placenta (18 -20). The importance of placental ADA was recently shown by genetically restoring ADA to the placentas of ADA-deficient fetuses (21), doing so prevented most of the purine metabolic disturbances as well as severe liver damage seen in ADA-deficient fetuses. The minigene used in this rescue also directed expression of ADA to the forestomach postnatally (19). The rescue of ADA-deficient fetuses provided the opportunity to investigate whether postnatal mice develop metabolic and immunologic features similar to those seen in ADA-deficient humans. Moreover, it enabled us to correlate metabolic changes in various tissues with ensuing phenotypes. In this study we report that mice homozygous for the null Ada allele, and rescued by expression of the ADA minigene, have no ADA expression outside of the gastrointestinal tract. This limited ADA expression is accompanied by pronounced tissue-specific metabolic disturbances including thymus-specific accumulations of deoxyadenosine and dATP, and inhibition of AdoHcy hydrolase in the thymus and spleen. In accordance with these tissue-specific disturbances, these mice developed an immune phenotype characterized by a significant reduction in the size of major lymphoid organs, a reduction in CD3-, CD4-, and CD8-positive cells in the spleen, and a decrease in lymphocyte responsiveness to mitogens. In some ways these mice have a metabolic and immunologic phenotype resembling that found in humans with a partial ADA deficiency. In this regard these mice represent the first genetic animal model with which to study the metabolic mechanisms responsible for immunodeficiency resulting from ADA deficiency.

EXPERIMENTAL PROCEDURES
Mice and Histology-Mice heterozygous for the null Ada allele (m1/ϩ) and hemizygous for the ADA minigene locus (Tg) were intercrossed to produce rescued mice that were homozygous for the null Ada allele (Tg, m1/m1) (21). Genotypes were determined by Southern blot analysis of genomic DNA obtained from tails at weaning (16,19). Organs from aged-matched Tg, m1/ϩ and Tg, m1/m1 adults (9 weeks old) were fixed in 4% paraformaldehyde, embedded in paraffin, sectioned, and stained with hematoxylin and eosin according to standard procedures.
ADA Enzymatic Assay-Aged-matched wild type (ϩ/ϩ) and Tg, m1/m1 adult mice were sacrificed and tissues dissected into ice-cold phosphate-buffered saline (PBS). Tissue extracts were generated and ADA enzymatic activity measured in high-speed supernatants under saturating substrate conditions using a spectrophotometric assay (19). The decrease in absorbance at 265 nm resulting from the deamination of adenosine to inosine was continuously monitored in a Beckman DU-50 spectrophotometer, and the rate of inosine production was calculated at linearity. Specific activities are presented as nanomoles of adenosine converted to inosine per min per mg of protein.
Zymogram Analysis-When tails were biopsied at weaning for the determination of genotypes, a drop of blood was also collected to monitor ADA enzymatic activity by zymogram analysis. One volume of homogenization buffer (19) was added, and cells were lysed by several rounds of freeze-thawing. Equal amounts of protein (20 g) were subjected to zymogram analysis to visualize ADA and purine nucleoside phosphorylase (PNP) enzymatic activities (18).
Analysis of Nucleosides and Nucleotides-Tissues from agedmatched Tg, m1/ϩ, and Tg, m1/m1 mice 9 -13 weeks of age were dissected into ice-cold PBS and then quick-frozen in liquid nitrogen for extraction and analysis of nucleosides (21,22). The HPLC system consisted of two Contrametric III pumps controlled by a GM 4000 gradient programmer, a constant wavelength UV monitor, and a CI-10B integrator (LDC Analytical). Separation was through a reversedphase Customsil ODS column (4.6 ϫ 254 mm) with a 20-mm ODS precolumn (Custom LC Inc.). The mobile phase was 0.2 M NH 4 H 2 PO 4 (pH 5.1) with a superimposed methanol gradient. Flow rate was 1 ml/min and the injection volume 200 l. Absorbance was continuously monitored at a wavelength of 254 nm, and peaks were identified and quantitated based on co-retention of known amounts of external standards (Sigma). Peaks of interest were verified by enzymatic shift assay. Whole blood was obtained by cardiac puncture, and serum and cellular components were separated by centrifugation at 2,600 rpm for 10 min at 4°C. Serum was processed for nucleoside levels, and packed erythrocytes were analyzed for nucleotides.
Nucleotides were extracted with 20 volumes of 60% methanol at Ϫ20°C overnight (21,23). Samples were filtered through Millex-GV 4 0.22 M membranes (Millipore) and stored at Ϫ20°C until analyzed by HPLC. The HPLC conditions were as described above except the ionpairing mobile phase consisted of 0.5 M NH 4 H 2 PO 4 (pH 6.5), 2 mM tetra-butylammonium hydroxide, and 7% acetonitrile, and the flow was isocratic.
littermates between 9 and 13 weeks of age and placed in ice-cold Dulbecco's phosphate-buffered saline containing 2% fetal bovine serum (PBS/BS). Cell suspensions were prepared by pressing organs through 60-mesh stainless steel screens followed by filtering through a Swinney filter to generate single cell suspensions. Erythrocytes in spleen cell suspensions were lysed by suspending the cell pellet in 1 ml of buffered ammonium chloride for 1-2 min, followed by addition of 10 ml of PBS/BS. Cells were washed and viability assessed using trypan blue exclusion. Viable cells were adjusted to 7.5-10 ϫ 10 6 cells/ml, and 100 l of cell suspension was dispensed in 12 ϫ 75-mm disposable glass tubes (Fisher) for reaction with various antibodies. Optimal concentrations of antibodies were diluted in 10-l volumes of PBS and distributed to appropriate tubes. Negative controls received 10 l of PBS/BS. Tubes were vortexed and incubated for 30 -40 min at 4°C. Cells were then washed twice with 300 l of PBS/BS. Two-color FACS analysis was conducted on 1 ϫ 10 4 viable leukocytes, and data were obtained on two-color contour profiles.
Proliferation Response Assay-Proliferation of spleen cells was determined by inducing blastogenesis with T and B cell mitogens and measuring incorporation of [ 3 H]thymidine into DNA. Single cell suspensions of spleens were prepared as described above. Spleen cells were suspended at a concentration of 10 6 cells/ml in RPMI 1640 medium, supplemented with 10% iron-supplemented calf serum (Sigma), antibiotics (penicillin-streptomycin 100 units/ml), and 5 ϫ 10 Ϫ5 2-mercaptoethanol. Cell suspensions (100 l) were dispensed in wells of roundbottomed 96-well plates (Corning) containing either 100 l of medium and optimal concentrations of the mitogens concanavalin A, 5 g/ml (Sigma), pwm, 2.5 g/ml (Life Technologies, Inc.), and bacterial lipopolysaccharide, 1.25 g/ml (Sigma), or medium alone. After incubation at 37°C in 5% CO 2 for 72 h, cultures were pulsed with [ 3 H]thymidine (1 Ci/well) and harvested 18 h later onto glass-filter discs using a PHD harvester. Discs were transferred to vials containing 1 ml of scintillation fluid, and counts per min were determined using a scintillation counter (Beckman). Stimulation index was determined by dividing mean cpm of experimental cultures by the mean cpm of control cultures.

ADA Enzymatic Activity in Rescued Mice Is Restricted to the
Gastrointestinal Tract-We have recently prevented perinatal lethality associated with ADA deficiency in mice by expressing ADA in the placenta of otherwise ADA-deficient fetuses (21). This genetic rescue provided mice which could be used to assess the metabolic and phenotypic consequences associated with ADA deficiency in mice. The DNA regulatory elements used to rescue ADA-deficient fetuses are known to target expression to the placenta prenatally and the forestomach postnatally (19). To confirm this expected pattern of ADA minigene expression, ADA enzymatic activity was measured in various tissues from adult rescued mice. Circulating ADA enzymatic activity was assayed by zymogram analysis of blood collected from mice during tail biopsies at weaning ( Fig. 2A). ADA was readily detected in wild type (ϩ/ϩ) and heterozygous (m1/ϩ) animals but was not detected in mice homozygous for the null Ada allele and carrying the ADA minigene locus (Tg, m1/m1). In addition to providing evidence that these animals were deficient in circulating ADA enzymatic activity, this procedure provided a sensitive and rapid means of identifying rescued mice. Homogenates from various tissues collected from ϩ/ϩ and Tg, m1/m1 adult mice were assayed for ADA enzymatic activity. As expected, high levels of ADA minigene expression, similar to those of native ADA, were detected in the forestomach of Tg, m1/m1 mice (Fig. 2B). Lower levels of ADA minigene expression were detected elsewhere in the gastrointestinal tract ( Fig.  2B) but no expression was detected in the thymus, spleen or other organs examined (Fig. 2C). Therefore, postnatal ADA expression from the minigene locus was restricted to the gastrointestinal tract, principally the forestomach.
Purine Metabolism Is Altered in Mice with Limited ADA Expression-ADA deficiency in humans and murine fetuses is associated with profound disturbances in purine metabolism (2,16) which are thought to provide the metabolic basis for immunodeficiency and perinatal lethality, respectively. Adenine nucleoside and nucleotide levels were measured in Tg, m1/m1 mice to determine the metabolic consequences associated with limited ADA expression in rescued mice. Acid-soluble extracts were made from various tissues from Tg, m1/ϩ and Tg, m1/m1 mice and adenine nucleosides were identified and quantitated using reversed phase HPLC. Representative HPLC profiles from thymuses are shown in Fig. 3 and indicate substantial increases in adenosine and deoxyadenosine in Tg, m1/m1 animals. Disturbances in purine metabolism were detected in all tissues examined (Fig. 4). Inosine levels were lower in the thymus, spleen, liver, and kidney of Tg, m1/m1 mice (Fig. 4A). Adenosine levels were elevated in all tissues examined ( 4B), with the greatest elevation occurring in the liver (9.4-fold) and thymus (7.3-fold). Deoxyadenosine accumulated in a thymus-specific manner, with levels increasing approximately 800-fold, from undetectable levels (Յ0.001 nmol/mg protein) in Tg, m1/ϩ thymuses, to 0.79 nmol/mg protein, in Tg, m1/m1 thymuses (Fig. 4C). These data demonstrate that there are profound disturbances in the levels of ADA substrates and products in mice with limited ADA expression. Moreover, there were thymus-specific elevations in the levels of deoxyadenosine found in Tg, m1/m1 mice, suggesting this lymphotoxic molecule may elicit a harmful effect on the immune system of these mice.
Deoxyadenosine is toxic to cells through mechanisms that include its phosphorylation to dATP (9) and subsequent inhibition of ribonucleotide reductase (Fig. 1) (6). The pronounced accumulation of deoxyadenosine in thymuses of Tg, m1/m1 mice prompted us to investigate adenine deoxynucleotide pools in these mice. Nucleotides were extracted from the thymus, spleen, liver, and erythrocytes of Tg, m1/ϩ and Tg, m1/m1 adult mice, and ATP and dATP were analyzed and quantitated by ion pairing reversed phase HPLC. ATP was readily detected in all samples (Table I), with slight elevations in ATP occurring in the thymus, spleen, and erythrocytes of Tg, m1/m1 mice. dATP was not detected at a lower limit of detection of Յ0.01 nmol/mg protein in any tissue examined except for the thymus of Tg, m1/m1 mice, where it increased over 50-fold. Consistent with these observations, dAMP, an intermediate in the phosphorylation of deoxyadenosine to dATP, was markedly increased in the thymus of Tg, m1/m1 mice (Table I, also see Fig.  3). The thymus and, to a lesser extent, the spleen were the only tissues in Tg, m1/ϩ mice in which dAMP was detected, albeit at low levels. These data suggest that deoxyadenosine accumulating in the thymus of Tg, m1/m1 mice is phosphorylated to dATP, providing compelling evidence for deoxyadenosine cytotoxicity in this organ.
AdoHcy Hydrolase Enzymatic Activity Is Inhibited in Tissues of Rescued Mice-Another metabolic consequence of ADA deficiency is the inhibition of AdoHcy hydrolase (11), which leads to the accumulation of AdoHcy and the disruption of cellular transmethylation reactions involving AdoMet ( Fig. 1) (12). To begin to assess the involvement of this pathway in adult mice with limited ADA expression, we measured AdoHcy hydrolase enzymatic activity in tissues of adult Tg, m1/ϩ and Tg, m1/m1 mice (Fig. 5). AdoHcy hydrolase enzymatic activity was detected in all tissues examined but in varying amounts (Fig. 5). AdoHcy enzymatic activity in the liver was at least 7-fold higher than the moderate levels measured in the small intestine, kidney, thymus, and spleen. The lowest levels of activity measured were in erythrocytes, brain, heart, and lung (data not shown). AdoHcy hydrolase enzymatic activity was substantially inhibited in the thymus and spleen of Tg, m1/m1 mice. Inhibition was also seen in the livers of these animals but to a lesser extent. There was no significant inhibition found in the small intestine or kidney of Tg, m1/m1 adult mice. These data suggest that AdoHcy metabolism is subject to lymphoid-specific disturbances in mice with limited ADA expression.
Mice with Limited ADA Expression Exhibit Lymphopenia and Partial Immunodeficiency-Thymus-specific accumula-  4. Levels of purine nucleosides and nucleotides in tissues from Tg, m1/؉ and Tg, m1/m1 mice. Tissues examined include thymus, spleen, liver, kidney, and serum. Nucleosides were extracted, analyzed, and quantitated by reversed phase HPLC (Fig. 3). Concentrations are given as nmol/mg protein for mean values Ϯ S.E.; n ϭ 3 for each genotype; ND, not detected at a lower limit of Յ0.001 nmol/mg protein. A, inosine; B, adenosine; C, deoxyadenosine. tions of deoxyadenosine and dATP, as well as the pronounced inhibition of AdoHcy hydrolase in thymuses and spleens, prompted us to investigate the status of the immune system in adult mice with limited ADA expression. All Tg, m1/m1 mice were viable and appeared in general good health. Upon gross examination of internal organs, the thymuses and spleens of Tg, m1/m1 adult mice were found to be significantly smaller in size than those of Tg, m1/ϩ littermates (Fig. 6, A and B). All other organ systems appeared normal including the liver which is the primary organ affected in ADA-deficient fetuses that die perinatally (16,17). Histological analysis of Tg, m1/m1 thymuses revealed a decrease in the size of the medullary region, which also appeared to contain fewer organized Hassal's corpuscles (Fig. 6, C and E). There was a decrease in the amount of red pulp observed in Tg, m1/m1 spleens (Fig. 6D, F), suggesting a possible effect on erythropoiesis. Consistent with this, fewer red blood cells were seen during the preparation of spleens for flow cytometry (data not shown). Consistent with the decrease in the size of thymuses and spleens, lymphoid cell counts were reduced from 40 million cells in Tg, m1/ϩ thymuses to 22 million in Tg, m1/m1 thymuses and reduced from 62 million cells in Tg, m1/ϩ spleens to 30 million in Tg, m1/m1 spleens (Fig. 7). Reduction in the size and cell number in these major lymphoid organs suggests that Tg, m1/m1 mice exhibit lymphopenia.
To evaluate leukocyte subpopulations, flow cytometry was performed on thymus and spleen cell populations from agedmatched Tg, m1/ϩ and Tg, m1/m1 mice using antibodies to the cell surface antigens, CD3, CD4, CD8, CD45R, GR-1, and CD11/b. Results from these analyses are shown in Table II.
Although the thymus of Tg, m1/m1 mice showed a large reduction in size and lymphoid cell number, there were no significant differences in the distribution of leukocyte subpopulations. There was, however, a slight reduction in CD4, CD8 double positive cells in Tg, m1/m1 thymuses. Leukocyte subpopulations were significantly altered in Tg, m1/m1 spleens. The most significant change was a decrease in spleen T cell populations. Cells positive for the T cell antigen CD3 were reduced from 45% in Tg, m1/ϩ spleens to 31% in Tg, m1/m1 spleens. Among the T cell populations, the greatest reduction was in CD4-positive cells (p Ͻ 0.001), with a decrease from 32% in Tg, m1/ϩ spleens to 17% in Tg, m1/m1 spleens. There was also a slight reduction in CD8-positive T cells. Significant increases in B cells (CD45Rpositive) from 45 to 56% were observed, as well as increases in granulocytes (GR-1-positive, 6 -9%) and macrophages (CD11/ b-positive, 8 -14%). These results indicate that Tg, m1/m1 mice exhibit alterations in leukocyte populations with the greatest effect being a CD4 lymphopenia. Stimulation of lymphocyte proliferation by mitogens is generally considered to reflect the function of immunocompetent cells. To begin to assess the status of immune function in Tg, m1/m1 mice, we monitored the stimulation of lymphocytes by T and B cell mitogens, concanavalin A for T cells, bacterial lipopolysaccharide for B cells, and pwm for T and B cells. Lymphoid cells from both Tg, m1/ϩ and Tg, m1/m1 spleens were stimulated by all three mitogens; however, the stimulation response in Tg, m1/m1 lymphoid cells was decreased compared with that seen in Tg, m1/ϩ lymphoid cells. The greatest difference in stimulation was by pwm (Fig. 8), with a stimulation index of 8.0 in Tg, m1/ϩ versus 4.1 in Tg, m1/m1. The 50% reduction in pwm stimulation may reflect the decrease in CD4-positive cells found in Tg, m1/m1 spleens. Collectively, these data indicate that Tg, m1/m1 mice with restricted ADA expression display moderate but significant levels of immunodeficiency. DISCUSSION ADA-deficient fetuses die perinatally (16,17), suggesting ADA is essential for fetal survival, but preventing our ability to assess the metabolic and immunologic consequences of ADA deficiency in adult mice. We were able to rescue these fetuses by restoring ADA to deficient placentas using an ADA minigene driven by Ada gene regulatory elements known to target expression to the placenta prenatally and the forestomach postnatally (21,19). In accordance with this, rescued adult mice homozygous for the null Ada allele and containing the ADA minigene locus did not express ADA in any tissue outside the gastrointestinal tract. Profound disturbances in purine metabolism were observed in these mice, with adenosine accumulating in all tissues examined, whereas deoxyadenosine and dATP FIG. 5. AdoHcy hydrolase enzymatic activity in tissues from Tg, m1/؉ and Tg, m1/m1 mice. The liver, small intestine (si), kidney, thymus, spleen, and erythrocytes (rbc) of aged matched adult Tg, m1/ϩ and Tg, m1/m1 mice were harvested and AdoHcy hydrolase enzymatic activity determined. Values are given as nanomoles of substrate converted per min per mg protein for means Ϯ S.E.; n ϭ 4; * , p Ͻ 0.01; ** , p Ͻ 0.005; *** , p Ͻ 0.001.

Consequences of Tissue-restricted ADA Expression in Mice
accumulated in a thymus-specific manner. AdoHcy hydrolase, a key enzyme in transmethylation reactions involving AdoMet, was markedly inhibited in the thymus, spleen, and to a lesser extent the liver but not in other tissues examined. The status of the immune system in these mice was examined, and a major reduction in the size of the thymus and spleen was found. In addition, changes in leukocyte subpopulations were observed with the greatest effect being a reduction in CD4-positive T lymphocytes found in the spleen. Mitogen stimulation assays showed a decrease in the responsiveness of T and B lymphocytes suggesting a partial immunodeficiency. These findings suggest that the lymphopenia and partial immunodeficiency in mice with limited ADA expression likely result from tissuespecific disturbances in purine metabolism. Our efforts have generated the first genetic animal model for studying the metabolic and immunologic consequences of ADA deficiency in postnatal mice. This has allowed us to investigate the metabolic impact of ADA deficiency in tissues not easily accessible in ADA-deficient humans. The most striking observation was the accumulation of deoxyadenosine and deoxyadenosine nucleotides in the thymus. Thymus-specific accumulation of deoxyadenosine supports the hypothesis that deoxyadenosine lymphotoxicity is involved in the immune phenotype found in rescued mice. Moreover, thymus-specific accumulation of dAMP and dATP provide evidence that deoxyadenosine is readily phosphorylated. This feature is consistent with in vitro studies suggesting that inhibition of ribonucleotide reductase by dATP is involved with phenotypes resulting from ADA deficiency (Fig. 1) (6 -10). Interestingly, dAMP was detected at low levels in the thymus of control animals. This may be a result of DNA breakdown from massive apoptosis which occurs in the thymus as a natural part of intrathymic T cell development and selection (24). Endogenous ADA levels in the thymus are high (20,25,26), most likely to handle this production of deoxyadenosine from dAMP. This is supported by the marked thymus-specific accumulation of deoxyadenosine and deoxynucleotides in mice lacking ADA in their thymuses. Recent studies have suggested that immature T lymphocytes are sensitive to the accumulation of deoxyadenosine through mechanisms involving p53-dependent apoptosis (27). It is possible that similar mechanisms are involved in the T cell lymphopenia observed in rescued mice and that T lymphocytes are depleted during intrathymic development. The animals generated here provide a valuable in vivo genetic model for learning FIG. 6. Reduction in lymphoid organ size. Spleen (A) and thymus (B) from 9-week-old Tg, m1/ϩ and Tg, m1/m1 littermates. A similar reduction in organ size was observed in all animals examined, n ϭ 26. Cross-sections through Tg, m1/ϩ (C) and Tg, m1/m1 (E) adult thymuses stained with hematoxylin and eosin. m, medullary region; c, cortical region; arrow, Hassal's corpuscle. Cross-sections through Tg, m1/ϩ (D) and Tg, m1/m1 (F) adult spleens stained with hematoxylin and eosin. w, white pulp; r, red pulp; bar ϭ 100 m. more about how disturbances in deoxynucleotide metabolism influence the growth and development of T and B lymphocytes.
Another proposed mechanism of immunodeficiency associated with ADA deficiency is the inhibition of methylation reactions involving AdoMet (11,12). The product of such methylation reactions is AdoHcy, which is hydrolyzed to adenosine and homocysteine by AdoHcy hydrolase (Fig. 1) (12,28). Inhibition of AdoHcy hydrolase can lead to the accumulation of AdoHcy, which functions as a competitive inhibitor of many transmethylation reactions critical to cellular function. Inhibition of AdoHcy hydrolase has been associated with ADA deficiency in humans (12) and in ADA-deficient perinatal mice (17). 2 The ability to examine individual tissues in mice with limited ADA expression revealed a correlation between the inhibition of AdoHcy hydrolase activity and the immune phenotype observed. Furthermore, inhibition of AdoHcy hydrolase enzymatic activity in the thymus together with the thymusspecific accumulations in deoxyadenosine are consistent with the inactivation of AdoHcy hydrolase by deoxyadenosine (11). However, this does not appear to be the case for AdoHcy hydrolase inhibition in other tissues that did not accumulate deoxyadenosine, such as the spleen and liver. There is evidence that AdoHcy hydrolase is inhibited by other nucleosides and nucleotides (29,30). In particular, patients with PNP and hypoxanthine-guanine phosphoribosyltransferase (HGPRT) deficiency show decreases in erythrocyte AdoHcy hydrolase enzymatic activity (30,31). Since deoxyadenosine is not elevated in these patients, AdoHcy hydrolase inhibition is likely due to other mechanisms. For example, inosine, whose levels can be elevated in the absence of both PNP and HGPRT enzymatic activity, has been shown to inactivate purified AdoHcy hydrolase, as well as in HGPRT-deficient lymphoblasts (30). Therefore, AdoHcy hydrolase inhibition in the spleen and liver of Tg, m1/m1 mice may result from purine metabolic disturbances other than deoxyadenosine accumulation. It is likely that these mice will provide a novel means of studying the mechanism(s) of AdoHcy hydrolase inactivation in animals and humans with genetic disorders of purine metabolism.
The other ADA substrate, adenosine, acts as an extracellular signal, engaging cell surface receptors to elicit an array of cellular functions (Fig. 1) (5). Adenosine has been shown to induce apoptosis in T lymphocytes in a receptor-mediated manner (32). Accumulation of adenosine was detected in all tissues of rescued mice examined; however, no obvious phenotype was observed in tissues outside the immune system, suggesting that increased adenosine concentrations are not overtly harmful. Although most evidence suggests deoxyadenosine is the primary lymphotoxic substrate in ADA deficiency, it cannot be ruled out that disruption of adenosine signaling in the thymus and spleen may be involved in the phenotype observed. Investigations into adenosine receptor distribution and engagement in these mice will help to clarify the involvement, if any, of adenosine signaling in the immune phenotype observed.
Disturbances in purine metabolism and subsequent immune phenotypes have been produced in mice treated with the ADA inhibitor 2Ј-deoxycoformycin (33)(34)(35)(36); however, the extent and duration of ADA inhibition in these pharmacological models is variable, making it difficult to accurately assess the metabolic and immunologic effects of ADA deficiency. Because of the genetic nature of the experiments conducted here, the impact of continuous ADA deficiency was made possible. Consequently, tissue-specific disturbances in deoxyadenosine were observed which correlate to partial immunodeficiency. The most striking effect on the immune system was the decrease in size and total lymphoid cell number in the thymus and spleen, suggesting lymphopenia. The medulla of ADA-deficient thymuses was considerably smaller in proportion to the cortex as compared with what is seen in normal mice. Given that the medulla harbors mature CD4 and CD8 single positive cells, one might expect a decrease in the percentage of this population ADA-deficient thymuses. This was not the case, however; there was a significant decrease in the percentages of CD4 and CD8 cells found in the spleen. The reason for this is not clear; however, it may reflect the inability of double positive cells to fully mature and emigrate out of the thymus. This feature has been noted pre-  viously in mice treated with 2Ј-deoxycoformycin (36). The thymus-specific accumulation of the lymphotoxic substrate deoxyadenosine in animals that lack ADA in all tissues outside the gastrointestinal tract does suggests that the origin of the T cell lymphopenia likely occurs during intrathymic development.
The majority of ADA-deficient humans harbor mutations severe enough to lead to a total loss of enzymatic activity resulting in severe combined immunodeficiency if not treated (2). These cases are accompanied by severe metabolic disturbances (2). A smaller population of ADA-deficient patients has been identified with late/delayed-onset of combined immunodeficiency disease (15). Often ADA mutations in these patients are less severe allowing for some ADA enzymatic activity and immune function. The decrease in the severity of immunodeficiency was associated with less severe elevation of deoxyadenosine nucleotides and less severe inhibition of AdoHcy hydrolase. This has lead to the hypothesis that the severity of ADAdeficient phenotypes relates to the overall capacity to eliminate deoxyadenosine (2,15). The rescued mice investigated here resemble these patients in that they have restricted levels of ADA expression, disturbances in purine metabolism, and partial immunodeficiency. It is not clear why the immunodeficiency was not more severe. However, one possibility is that the expression of ADA in the gastrointestinal tract of these mice minimizes the metabolic consequences of these otherwise ADAdeficient mice. Removal of ADA from the gastrointestinal tract may result in a more severe metabolic disturbance and immunodeficiency. Current efforts are underway to rescue ADAdeficient fetuses using DNA regulatory elements that target expression of an ADA minigene to the placenta only thus eliminating postnatal expression in the gastrointestinal tract. Such animals will completely lack ADA when born and may develop a more severe immune phenotype than that reported here.
In conclusion, mice generated by the genetic rescue of ADAdeficient fetuses exhibit metabolic and immunologic features relevant to those seen in ADA-deficient humans. The ability to monitor metabolic changes in tissues not easily accessible in ADA-deficient humans has enabled us to obtain in vivo data that strengthen the hypothesis that deoxyadenosine accumulation in the thymus is responsible for the immune phenotype associated with ADA deficiency. Furthermore, we provide in vivo evidence that disturbances in both deoxynucleotide metabolism and AdoMet-mediated transmethylation reactions are involved. These mice will be a useful genetic model for assessing the relative involvement of each of these pathways in the T and B cell lymphopenia associated with ADA deficiency.