Telomere length regulation during postnatal development and ageing in Mus spretus
Telomere length regulation during postnatal development and ageing in Mus spretus Gina M. Coviello-McLaughlin and Karen R. Prowse*
Geron Corporation, 200 Constitution Drive, Menlo Park, CA 94025, USAReceived April 23, 1997; Revised and Accepted June 11, 1997
ABSTRACT
Telomere shortening has been causally implicated in replicative senescence in humans. To examine the relationship between telomere length and ageing in mice, we have utilized Mus spretus as a model species because it has telomere lengths of approximately the same length as humans. Telomere length and telomerase were analyzed from liver, kidney, spleen, brain and testis from >180 M.spretus male and female mice of different ages. Although telomere lengths for each tissue were heterogeneous, significant changes in telomere lengths were found in spleen and brain, but not in liver, testis or kidney. Telomerase activity was abundant in liver and testis, but weak to non-detectable in spleen, kidney and brain. Gender differences in mean terminal restriction fragment length were discovered in tissues from M.spretus and from M.spretus * C57BL/6 F1 mice, in which a M.spretus-sized telomeric smear could be measured. The comparison of the rank order of tissue telomere lengths within individual M.spretus showed that certain tissues tended to be longer than the others, and this ranking also extended to tissues of the M.spretus*C57BL/6 F1 mice. These data suggest that telomere lengths within individual tissues are regulated independently and are genetically controlled.
INTRODUCTION
Telomeres, the natural ends of linear eukaryotic chromosomes, consist of evolutionarily conserved, repetitive DNA and its associated specific DNA binding proteins. In most vertebrates, telomeric DNA consists of a short G-rich sequence, 5'-TTAGGG-3', repeated hundreds to thousands of times at the ends of all chromosomes (1 ). Telomeres function to protect the loss of vital DNA sequences through nuclease degradation or illicit recombination (reviewed in 2 ). They have been shown to bind to the nuclear matrix (3 ) and may participate in chromosome pairing during meiosis (4 ). Telomeres have also been implicated in the control of cell immortalization (5 ,6 ) and cellular senescence (7 ,8 ). Cellular senescence was first described by Hayflick and Moorehead (9 ) as the point at which a cessation of cell growth occurs, concomitant with a lack of response to normal growth stimuli, in a population of human fibroblasts in culture. The number of possible cell doublings decreased as a function of donor age, presumably reflecting the replicative history of the cells in vivo. This limited cell division capacity has since been observed in numerous other somatic cell types, and has led to the idea of a cellular mitotic clock, which measures cell divisions rather than chronological age. Evidence has accumulated which suggests that telomeres play a role in the clocking mechanism.
In humans, telomeres are known to shorten during replicative ageing in various types of somatic cells at rates ranging from 40 to 200 bp/division depending on the cell type (5 ,10 -12 ). This shortening of telomeres was predicted based upon the end-replication problem of eukaryotic linear chromosomes (13 ,14 ). DNA polymerases require an RNA primer to initiate replication and synthesize DNA unidirectionally in the 5' to 3' direction. The removal of the RNA after each round of DNA replication results in a loss of sequence at the 5' end of the daughter strands. The telomere hypothesis of cellular senescence (15 ) proposes that when a critically short length of telomeric sequence remains on one or a few chromosomes in normal somatic cells, the cells cease to divide to prevent self-destruction. Without a mechanism to overcome this under replication and maintain a complete set of chromosomes, cellular and ultimately organismal viability would be impaired. In germline, stem cells and most immortal and tumor cells, a unique DNA polymerase, telomerase, acts to compensate for incomplete replication of chromosomes, thus permitting continued cell division (reviewed in 16 ). Telomerase, a ribonuclear protein complex, synthesizes telomeric DNA de novo onto the 3' end of the parental G-rich strand using its integral RNA as a telomeric sequence template (17 -19 ). The newly synthesized telomeric DNA provides a template to allow the conventional DNA polymerase complex to synthesize a complete daughter strand.
The biochemical properties of telomerase and regulation of telomere length between human and mice differ. Partially purified mouse telomerase extracts adds predominately only one repeat onto a telomeric primer in vitro under conditions where the human enzyme adds hundreds of repeats (20 ). The telomerase RNA components between human and mice share only 65% sequence identity and differ in their template regions (19 ,21 ). A recently identified, non-catalytic telomerase-associated protein shows 75% amino acid identity between humans and mice in the ORF (22 ). The terminal restriction fragment (TRF) lengths, comprising both telomeric and non-telomeric sequences, also differ between human and mouse DNA. Tissues from humans have mean TRF lengths that range from 5 to 15 kb (10 ,23 ), whereas tissue TRFs from most inbred strains of Mus musculus are >50 kb (24 ,25 ). Telomeric sequence binding proteins identified in mouse and human cells share an overall homology of 67% indicating perhaps a rapid evolution of this gene as well (26 ). In humans, most normal somatic cells do not have detectable telomerase activity (5 ,27 ) and telomeric DNA is lost during replicative cell division in culture (10 ) and during ageing in vivo (11 ,28 ). However, in mice, many normal somatic cells do express detectable telomerase activity (29 ,30 ), but the effect of this activity on telomere length during ageing was largely unknown.
To examine the relationship between ageing and telomere length in mice further, a suitable mouse model is necessary. Because of the long TRF lengths of M.musculus, detection of any changes in TRF length with age is difficult by conventional gel separation techniques. In contrast, Mus spretus and Mus caroli, both wild-derived species, have been shown to have significantly shorter mean TRF lengths (5-25 kb) than M.musculus strains (25 ). In a previous study, we examined seven additional wild-derived mouse species, all of which had shorter telomere lengths than M.musculus strains (29 ). We chose to follow telomere length regulation in M.spretus because of its readiness to interbreed with laboratory strains of M.musculus, thus providing further opportunities for genetic studies of telomere length.
Previously, we reported differences in telomere length between tissues within a few individual M.spretus (29 ). Due to the limited number of mice sampled in the study, it was not possible to detect changes in tissue telomere length with respect to ageing in M.spretus. In this study, we have examined telomere lengths and telomerase activity in five tissues in a large population of randomly bred male and female M.spretus at different ages. Statistically significant TRF length changes with age were found in brain (0-4 months) and spleen (0-30 months), but not in liver, testis and kidney (0-30 months). A significant difference between male and female tissue TRF lengths in M.spretus was also discovered. In addition, a M.spretus-like telomeric smear was detected in F1 mice from M.spretus * C57BL/6 F1 crosses, which displayed similar tissue and gender TRF length differences. Abundant telomerase activity was present in liver and testis, while weak or no activity was detected in spleen, kidney and brain. These results are discussed in relation to the role that telomere length and telomerase may play in replicative ageing in mice.
MATERIALS AND METHODS
Mice
Ten male and 10 femaleM.spretus were obtained from Jackson Laboratory (Bar Harbor, ME) and maintained as breeding pairs. The colony was expanded by using a random breeding system among the offspring over a 2.5 year period in a licensed research animal facility. Mice were maintained in VCLtm ventilated caging units (Lab Products, Maywood, NJ) in a clean room environment and were sustained on a rodent breeder diet (Harlan Teklad #8760) and acidified water, pH 3.0, + 0.4 [mu]g/ml Vitamin K. Several female C57BL/6 mice were also purchased from the Jackson Laboratory, crossed to male M.spretus, and maintained as described above. The ageing study was conducted according to an IACUC (Institutional Animal Care and Use Committee) approved protocol. Mice were euthanized with ether at the different time points and samples of kidney, spleen, liver, brain, testes and serum were collected and frozen on dry ice. Some of the M.spretus brain samples were dissected, separating out the cerebellum, hippocampus and the subventricular zone of the basal ganglia and frozen on dry ice.
Isolation and restriction enzyme digestion of genomic DNA
Genomic DNA was prepared from mouse tissue samples according to a standard protocol (29 ). Briefly, frozen tissues were rapidly homogenized (PowerGen 35, 5 * 95 mm generator probe; Fisher Scientific) in DNA extraction buffer (100 mM NaCl, 10 mM Tris, 25 mM EDTA, pH 8, 0.5% SDS, 0.1 mg/ml Proteinase K). The homogenate was then treated as described previously. DNA was dissolved in 1* TE and heated at 50-55oC for 1-5 h. A fluorimeter was utilized to quantitate the isolated genomic DNA. The integrity of the undigested DNA was analyzed by gel electrophoresis. Degraded, uncut DNA appeared as a smear extending from limit of mobility to the bottom of the gel, while intact DNA appeared as a band at limit of mobility. An aliquot (3-5 [mu]g) of DNA was digested with HinFI/RsaI(2 U each/[mu]g DNA) and DNase-free RNase at 37oC for 16 h. The digests were monitored for completeness by gel electrophoresis. Incomplete digests appeared as a band or smear at the top of the gel, while the complete digests appeared as a smear at the bottom of the gel.
TRF analysis
Aliquots (1-2 [mu]g) of digested DNA were separated on a 0.5% agarose gel (20 * 25 cm) in 1* TBE for a total of 800-900 V h. The gels were dried for 20-25 min at 60oC, denatured (0.5 M NaOH, 1.5 M NaCl) for 8 min and neutralized (0.5 M Tris pH 7, 1.5 M NaCl) for 4 min. Gels were prehybridized in 30 ml of a standard hybridization solution (5* Denhardt's solution, 5* SSC, 10 mM Na2HPO4, 1 mM Na4P2O7) at 37oC for 1-4 h. An aliquot (0.25 [mu]g) of a single-stranded telomeric oligonucleotide, (TTAGGG)3, was end-labeled with 50 [mu]Ci of [[gamma]-32P]ATP and 10 U of T4 polynucleotide kinase and added to the prehybridization buffer. The gels were incubated at 37oC for 16 h, and washed with 0.5* SSC at 37oC (3 * 10 min). To determine the TRF length, a Phosphorimager (Molecular Dynamics) was used to quantitate the position and strength of the radioactive signal in each of the lanes, as described (7 ,29 ).
Tissue extracts and telomerase assays
Mus spretus tissues were harvested and frozen as described above. Samples were kept cold and homogenized with a motorized disposable pestle in CHAPS lysis buffer (10 mM Tris-HCl, pH 7.5, 1 mM MgCl2, 1 mM EGTA, 0.1 mM phenlymethylsulfonyl fluoride, 5 mM [beta]-mercaptoethanol, 0.5% CHAPS, 10% glycerol) using 20 [mu]l of lysis buffer/4-10 mg of tissue and placed on ice for 30 min. Samples were microcentrifuged at 12 000 g for 30 min at 4oC. A Coomassie assay (Pierce) was used to quantitate the amount of protein in each extract and TRAP assays were performed using 2 [mu]g of protein as described previously (27 ). Samples with detectable signal above the background in the negative control lanes were normalized to an internal control PCR product and related to a known amount of DNA standard. The quantitation was expressed as total product generated.
RESULTS
Telomere length changes in M.spretus during ageing
In this study, we have analyzed a population of M.spretus for telomere length in multiple tissues throughout postnatal development and ageing. Tissues from 109 female and 73 male M.spretus were collected at different times over a 2.5 year period, and genomic DNA was isolated from kidney, liver, spleen, testes and brain. The mean TRF length of each sample was determined using a phosphorimager as described in the Materials and Methods (7 ,29 ). Figure 1 shows a representative tissue gel of three male and three female M.spretus of different ages. As expected, the signal was distributed over a wide size range (4-15 kb) representing both strongly hybridizing smears and weakly hybridizing, distinct bands of DNA. The smears were determined to be telomeric by sensitivity to Bal31 exonuclease digestion, while most distinct bands represented internal tracts of TTAGGG (data not shown). The mean TRF lengths for each tissue were varied among individuals. For example, the liver DNAs in Figure 1 show a difference in mean TRF length between each pair of same-aged mice, as well as among the different ages.
Comparison of telomere lengths within individual M.spretus
As different tissues may differ in their turnover rate in vivo, we compared telomere lengths of different tissues within individual mice. Fifty mice (26 male and 24 female), ranging in age from 2 to 26 months postpartum, were used for the analysis; only mice with TRF length data for all tissues were included, and the data were not separated initially by age of the mouse. We first calculated the average TRF length for each tissue and compared those values (Table 2 ). On average, testis and brain TRFs were the longest and spleen TRFs were the shortest. However, because the heterogeneity in TRF lengths between mice of the same (as well as different) ages made absolute length comparisons difficult, we also rank ordered the tissues within each mouse from longest (1) to shortest (4 or 5), calculated the average rank for each tissue for males and females, and performed pairwise t-test analyses on the ranked data (Table 2 ). Although an exact order of tissues could not be definitively determined, significant differences between tissue TRFs within individual mice in the population were found. In males, testis and brain TRFs were statistically different from each other and all other tissues examined. No differences were found between liver, kidney and spleen. In females, brain TRFs were also significantly different from kidney and spleen, and spleen was different from liver. When we examined tissue TRF order with respect to age by dividing male and female mice into two roughly equal-sized age groups, similar results were obtained (data not shown). These data suggest that although there were variations in order among individual mice, telomere length maintenance within the different tissues and relative to other tissues may be genetically pre-programmed. If the relative lengths within each tissue are controlled genetically, controlling factors may be the presence or absence of telomerase activity in each tissue and/or the degree of cell turnover.
a1 = Longest TRF. Values for pairwise two-tailed t-tests on M.spretus tissue rankings are shown. Values in italics are at least significant at the 0.05 level.
Total number, percent and ages of telomerase positive tissue samples are shown. Ages of telomerase negative samples are also given.
Telomerase activity in M.spretus with age
Telomere lengths reflect gender differences
TRF lengths were plotted versus age for each tissue, and linear regression analyses were performed for males, females and the total population (Figs 2 -4 , Table 1 ). When male and female TRF lengths were compared for each of the tissues, significant gender differences were observed (Table 4 ). The regression line for females in both liver and kidney was ~+0.5 to +1.0 kb offset from the male regression line. In the spleen, male and female TRF lengths each showed significant decreases (P = 0.02 and 0.00016) at approximately the same rate (46 and 42 bp/month). Again, a significant gender difference (P = 0.00001) was observed between the regression lines of the two populations (Table 4 ), with female TRFs being ~0.5 kb longer. Gender differences were also present in brains of M.spretus >= 5 months old. When males and females in the >= 5 month group were compared for telomere lengths (Fig. 4 C), female TRFs again showed a significantly longer mean TRF length (+0.5 kb, P = 0.0075). A limited number of male mice prevented the analysis of gender effects in the 0-4 month group.
Two other lines of evidence support that telomere length is influenced by gender. First, pairwise analyses of the ranked TRF length data yielded differences in significance between tissues with respect to gender. Second, in a limited analysis of brain and spleen DNAs from four male and four female M.spretus * C57BL/6 F1 mice, preliminary data showed two peaks of hybridization signal in the F1 samples. Peak 1 (>30 kb) appears to correspond to the C57BL/6 parent while peak 2 (14-20 kb) may derive from the M.spretus parent (data not shown). A statistically significant difference in TRF peak 2 between genders in both tissues was observed (Table 4 ). In addition, the average spleen TRF lengths for both male and female F1 mice (15.21 and 18.94, respectively) were shorter than those for brain (16.07 and 19.87). Both these results are consistent with the data from M.spretus tissue TRF length analysis described above, and suggest that peak 2 in the F1 mice derives from the paternal M.spretus chromosomes.
Significant gender differences in tissue TRF length
Tissue
Student t-testa Males versus females (P value)
Kidney
0.0002
Liver
0.00002
Spleen
M.spretus
0.00001
spr * BL/6
0.02
Brain
M.spretus
all ages
0.87
>= 5 months
0.0075
spr * BL/6
0.014
aTwo-tailed t-test. Values in italics are statistically significant. spr * BL/6, F1 mouse from M.spretus * C57BL/6 cross.
DISCUSSION
Differences in telomere length and telomerase activity between and within mouse and human cells and tissues have previously been reported (20 ,29 ,30 ,33 ). The TRF lengths of mouse species can differ quite dramatically, from >50 kb in M.musculus laboratory strains (24 ,25 ) to the 5-15 kb in M.spretus and other wild-derived mice (25 ,29 ). However, the TRF lengths in human tissues are considerably shorter (5-15 kb) and less variable than the hypervariable M.musculus. In addition, unlike most human somatic cells, telomerase is expressed in many mouse somatic tissues (29 ,30 ). Furthermore, telomere binding proteins, and telomerase activity, RNA and one associated protein component show different sequences and properties between mouse and human (20 ,22 ,26 ,34 ). Given these distinctions, it is not unlikely that telomere length regulation in mouse may involve some altered or additional pathways from telomere length regulation in humans. In this study, we have utilized the wild-derived mouse species, M.spretus, to examine the relationship between cell ageing and telomere length regulation in the mouse.
TRF lengths in the same tissue types between many individual mice of different ages in a population were compared. In this analysis, we determined whether telomere length regulation in mice followed the telomere hypothesis of cell senescence proposed for human cells (15 ). The telomere hypothesis proposed that telomeres will shorten during replicative cell division due to the inability of normal DNA polymerases to completely replicate the ends of linear chromosomes in those cells that lack sufficient telomerase activity. Once a critical telomere length is reached, the shortened telomere(s) would signal the cell to stop dividing, perhaps using DNA damage pathways. In cells that have sufficient telomerase activity, telomere length would be maintained from one generation to the next, and allow for continued cell division. Some of our results from mice support the telomere hypothesis of cell ageing, while other results may support alternative or additional explanations for maintaining telomere length.
Mus spretus tissues differed in TRF lengths and telomerase activity with age. None of the TRF lengths in any tissue or at any age were <5 kb, which is in agreement with the TRF length found in human cells at senescence (8 ). Liver and kidney samples showed no statistical difference in TRF length with age. Although data regarding accurate cell turnover is difficult to obtain, some reports suggest that both of these tissues show little cell division in normal tissues in adult mice (35 ,36 ), so no changes in TRF length might be expected. However, a difference in telomerase activity between the two tissues is apparent: telomerase was detected in 75% of liver samples but only 12.5% of kidney samples from all ages. To account for this discrepancy, the telomere hypothesis would propose that telomerase activity is present in these non-dividing, adult liver cells to maintain telomere lengths in the event of liver damage. This activity could allow the necessary cell divisions to promote regeneration of damaged tissue to occur. Contrary to the liver and kidney data, the spleen showed distinctive TRF length changes with age. A portion of the spleen contains different types of proliferating cells that continue to divide throughout the lifespan (37 ). The low level of telomerase activity detected in 25% of spleen samples probably derived from the small population of self-renewing progenitor cells as previous work (38 ) has shown. For our purposes in this study, whole spleens were homogenized into extracts without separation of the progenitor population. The decrease in whole spleen TRF length with age probably reflects the non-self-renewing cell proliferation occurring in the spleen, in agreement with the telomere hypothesis.
No significant change in testis TRF length with age was observed in the population as a whole. However, in separate age group analysis, testis TRFs show a statistically significant increase with age up to 12 months. Interestingly, an increase in the TRF length of mature human sperm with age has also been detected (7 ). This increase in mouse TRF lengths correlates with the presence of telomerase activity in the testis. Between 4 and 8 weeks postpartum, the testis is ~80-90% spermatogonial cells and the first mature spermatozoa are produced (39 ). Telomerase activity had previously been detected by the conventional telomerase assay in M.musculus between 4 and 6 weeks postpartum (29 ), suggesting that telomerase is present at some stage of spermatogonial development. Likewise, we detected telomerase activity by TRAP assay in the testis of all M.spretus 4 weeks or older, but not in two 4 day old mice. Testis telomerase expression is therefore consistent with the ability to maintain or increase telomere length in the dividing germ cells.
The testis, spleen and kidney appear to support the telomere hypothesis in mice and the liver data could also speculatively support the hypothesis. Telomere length dynamics in brain, however, do not support the telomere hypothesis of ageing. Brain TRFs show a biphasic regulation of length, which appears unrelated to cell turnover or telomerase. TRF lengths in brain decrease at a rate 10-fold faster from 0 to 4 months than in mice >= 5 months, though the decrease in the older group is non-significant. The significant TRF decrease in the younger group does not seem to arise from rapid cell proliferation because in the >2 week postpartum brain, cell division seems limited to a small population of stem cells within the hippocampus, cerebellum and subventricular zone of the basal ganglia (31 ,32 ,40 ). In the younger age group, the range of mean TRF lengths is quite homogeneous, while in the older group it is very heterogeneous. Brain telomere lengths also increase substantially on average at 5 months and are maintained at the new length thereafter. However, telomerase activity seems not to be involved because telomerase is not detected in total brain samples at any age. In addition, no activity was detected in any of the individual regions thought to have continued cell proliferation, although it is possible that the isolated stem cells may contain telomerase activity which may be masked by the bulk of the tissue. The significant decrease in TRF length in the brain, followed by a significant increase at 5 months may therefore involve alternate mechanisms of telomere-length modulation.
The finding of gender differences in tissue TRF lengths in all four tissues examined is one line of evidence which points to a genetic component of mouse telomere length regulation in tissues. Male M.spretus had TRF lengths that were on average 0.5-1 kb shorter TRFs than females in every tissue examined. Male M.spretus * C57BL/6 F1 mice also had shorter peak 2 TRFs than the female F1 mice, although in these mice the difference was ~3.7 kb rather than 0.5-1 kb. Although we have not observed a consistent weight difference between adult male and female M.spretus, a growth rate or metabolic distinction between genders may be a possible explanation for the TRF length difference. Gender TRF length differences in at least one other species has been found. Rhesus monkey males displayed lymphocyte TRF lengths that were, on average, 2.5 kb longer than those of female monkeys (unpublished data). However, in human lymphocytes, no significant difference between males and females was detected (11 ). This finding of a gender effect on telomere length in mice could be another difference between mice and humans or it could simply be that not enough human tissues have been examined to detect a difference. Further telomere regulation and metabolism studies in humans and mice may provide a possible explanation.
Significant differences between tissue TRF lengths within individual mice also support a genetic component of telomere length regulation. When tissue TRFs within individual M.spretus were compared, specific tissues tended to have longer or shorter TRF lengths compared to other tissues. The finding of TRF length differences between tissues suggests that TRF lengths of each tissue are regulated independently. This tendency is also observed in the preliminary data from M.spretus * C57BL/6 F1 mice since both the gender and size difference were maintained. To follow this phenomena further, a significant number of F1 mice as well as subsequent generations and backcrossed animals should be examined. One other line of evidence which suggests a genetic component is the fact that the American colonies of M.spretus have shorter TRF lengths than their European counterparts (41 ). The telomeres of European M.spretus are similar in size to M.musculus strains. The American colonies presumably derived from a small population of the European M.spretus in which perhaps one or more changes occurred which affected telomere length regulation (41 ).
Although most tissue data in this study support the telomere hypothesis of cell ageing, the results from the brain data suggest that telomere lengths may decrease due to factors other than the end-replication problem during cell division, and that alternate pathways in addition to those utilizing telomerase may be involved in maintaining and regulating telomere length in the mouse. Both environmental as well as genetic factors could be involved. Recently, hyperoxic conditions have been shown to rapidly shorten telomeres in cultured cells without cell turnover (42 ). The mechanism for this rapid decrease is thought to be due to free radical-induced single strand breaks in the telomeric DNA. Endogenous factors, including those involved in telomere structure, DNA repair and cell cycle regulation, also very likely play a role in regulating telomere length in the mouse. For example, differential regulation of the mammalian telomere-repeat binding factor (TRF1, 26 ) could be one mechanism involved in setting the telomere length in distinct tissues by regulating the availability of telomeric sequences to modifying factors. Likewise, perhaps, DNA repair enzymes, such as DNA polymerase [beta] which has a high activity level in the brain (43 ), or other DNA replication factors play a direct or indirect regulatory role in the telomere dynamics of the mouse brain. Telomere length regulation during ageing in M.spretus tissues may be a complex dynamic process involving not only the coordinated regulation of telomerase activity and cell cycle controls, but also additional, perhaps tissue-specific, factors. Further studies should yield important clues to understanding telomere length regulation in mouse.
ACKNOWLEDGEMENTS
We would like to thank Shari Starr for animal and protocol support, Jeff Roberts for veterinary care, Clive Svendsen for helpful discussions on mouse brain development, Junko Aimi and Mike Kozlowski for mouse brain dissection, Fred Wu for providing the TRAP kit and consultation, and Choy-Pik Chiu and Calvin Harley for critical review of this manuscript.
REFERENCES
1 Meyne,J., Ratliff,R.L. and Moyzis,R.K. (1989) Proc. Natl. Acad. Sci. USA86,7049-7053.MEDLINE Abstract
26 Broccoli,D., Chong,L., Oelmann,S., Fernald,A.A., Marziliano,N., van Steensel,B., Kipling,D., Le Beau,M.M. and de Lange,T. (1997) Hum. Mol. Genet.6,69-76.MEDLINE Abstract
37 Forni,L. (1988) Ann. Inst. Pasteur Immunol.139,257-266.
38 Morrison,S.J., Prowse,K.R., Ho,P. and Weissman,I.L. (1996) Immunity5,207-216.
39 Whittingham,D.G. and Wood,M.J. (1982) In Foster,H.L., Small,J.D. and Fox,J.G. (eds), Reproductive Physiology. Academic Press, New York, pp. 139-142.
40 Wood,K., Dipasquale,B. and Youle,R. (1993) Neuron11,621-632.MEDLINE Abstract
*To whom correspondence should be addressed at present address: Cell Engineering Facility, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands. Tel: +31 50 363 7290; Fax: +31 50 363 4165; Email: kprowse@aol.com
