Jorde-lab.genetics.utah.edu

Female geneflowstratifies Hindu castesScientists have long been interested in First, there may not be enough Y-chromo- understanding how social processes some STRs to reveal a meaningful pattern, modulate evolutionary forces1. A good although a comparison of a distance matrix example of this is the intensively studied based on the Y-chromosome SNPs with the Hindu caste system, which governs the mat- matrix based on STRs yields a sig ing practices of nearly one-sixth of the positive correlation (r 5 0.68, P , 0.0025).
world's population2,3. However, there is This high degree of consistency would not be controversy concerning the effect of social expected if the STR distances were due to stratification on the genetic structure of insufficient data.
caste communities. Here we show that dif- Second, the lack of correlation between ferences in social rank between castes corre- social rank and Y-chromosome distance spond to mitochondrial DNA (mtDNA) might be produced by random male move- distances between castes but not genetic Figure 1 Neighbour-joining networks7 of genetic dis- ment between castes. This explanation is distances estimated from Y-chromosome tances among caste communities. a, Network esti-
unlikely because the Y-chromosome data mated from 411 base pairs of hypervariable region 1 show greater differentiation between groups Hindu populations are stratified into of mtDNA8. Distances between upper castes (Brah- than the mtDNA data: the proportion of five varna, each of which is associated with min, n 5 41; Vysya, n 5 10; and Kshatriya, n 5 10), genetic variation attributable to differences a ‘status'. Ranked from lowest to highest sta- middle castes (Yadava, n 5 48; and Kapu, n 5 52), between groups for Y-chromosome STRs is tus, these are Panchama, Sudra, Vysya, and lower castes (Relli, n 5 20; Mala, n 5 25; and 0.10 (P , 0.001), whereas that based on Kshatriya and Brahmin4. There are approxi- Madiga, n 5 28) are correlated with social rank. Pair- mtDNA data is only 0.015 (P , 0.002).
mately 2,000 castes in contemporary India, wise estimates of the proportion of genetic variation This result is consistent with females having roughly grouped into these varna. Religious attributable to differences between groups indicate greater intercaste gene flow than males, and rites, access to material resources and edu- that the genetic distance between upper and lower agrees with historical evidence that males cation, and occupational specialization are castes is fourfold higher than between upper and were less likely to change castes.
directly related to caste status. Marriages middle castes (P , 0.001). A neighbour-joining net- A third possible explanation is that the between individuals of equal status are pre- work of unique haplotypes reveals a star-like pat- discordance is caused by differences in ferred. Matings between a man from a tern (data not shown) with a few short central mutation rates between the Y-chromosome higher varna and a woman from a lower branches linking nodes from which tufts of long and mtDNA systems6. However, the greatest varna are permissible under certain circum- branches emerge. Lineages from different castes observed difference in mutation rates is stances, in which case the offspring tend to are scattered throughout the network, consistent between the Y-chromosome STRs and the attain a status similar to that of their father.
with gene flow between communities or the sorting Y-chromosome SNPs, yet these two data In contrast, marriage of a woman from a of lineages before the separation of castes. Analy- sets are highly concordant.
higher varna to a man of a lower varna is ses limited to transversions or excluding rapidly The remaining explanation for the dis- strongly discouraged. This suggests that mutating sites within mtDNA HVS1 (ref. 9) do not cordance, and the one most consistent with women have limited but upward social substantially alter these results (data not shown). b,
the historical data, is that systematic female mobility, whereas men have very little4.
Network of genetic distances among caste commu- gene flow among castes produced a correla- To test the effects of this mating system nities estimated from Y-chromosome STRs (DYS19, tion between social rank and mtDNA dis- on genetic variation, we analysed mtDNA DYS288, DYS388, DYS389A, DYS389B, DYS390 and tances. A relative lack of male gene flow and Y-chromosome variation in 250 indi- DYS392) and SNPs (SRY10831, SY57 and RPS4Y)10.
resulted in no correlation between social viduals from 12 Telegu-speaking caste pop- Upper, middle and lower castes do not cluster rank and Y-chromosome distances, and ulations from northeastern Andhra Pradesh together, and distances between upper, middle and supports the idea that the pattern of Y- in southern India. Castes and varna were lower castes are not correlated with social rank.
chromosome variation is largely the result ranked by status into upper (Brahmin, of mutation and genetic drift. This indicates Kshatriya and Vysya), middle (Kapu, Analysis of Y-chromosome data revealed that mobility between castes of different Yadava, Jalari and Wadabahija), and lower that seven short-tandem repeats (STRs) and status has been higher for females than (Relli, Madiga and Mala) groups.
three single-nucleotide polymorphisms males, as predicted from ethnographic We identified 182 unique mtDNA haplo- (SNPs) produced 107 unique haplotypes in records. We therefore conclude that genetic types, four of which are shared among 194 individuals for whom the data were stratification of the Hindu caste system is upper, middle and lower castes, seven complete. For the STR data, the genetic dis- driven by the social mobility of women.
between lower and middle castes, and three tance between upper and lower castes Michael J. Bamshad*, W. Scott Watkins†,
between middle and upper castes. No haplo- (0.0054) is similar to that between the Mary E. Dixon†, Lynn B. Jorde†,
types are shared exclusively between upper upper and middle castes (0.0062), and larger B. Bhaskara Rao‡, J. M. Naidu‡,
and lower castes. This suggests that haplo- than that between middle and lower castes B. V. Ravi Prasad‡, Arani Rasanayagam§,
type sharing between castes is limited by (0.0005). Thus, caste status is not associ- Mike F. Hammer§
social rank. The genetic distance between ated with Y-chromosome genetic distance Departments of *Pediatrics and †Human Genetics, upper and lower castes (0.00045) is 1.5 times Eccles Institute of Human Genetics, greater than between the upper and middle A statistical comparison of the genetic University of Utah, (0.00024) or middle and lower castes distances based on mtDNA and Y-chromo- Salt Lake City, Utah 84112, USA (0.00030). Thus, mtDNA distances between some data reveals a low, non-significant castes of different status are stratified correlation (r 5 0.14, P . 0.3) by the Man- ‡Department of Anthropology, according to social rank (Fig. 1a). This is tel permutation test5. There are several pos- Andhra University, Visakhaptnam, consistent with higher levels of female gene sible explanations for this lack of Andhra Pradesh, India flow between more closely ranked castes.
§Laboratory of Molecular Systematics and Nature Macmillan Publishers Ltd 1998
NATURE VOL 395 15 OCTOBER 1998 www.nature.com Evolution, University of Arizona, showed that indomethacin administration Tucson, Arizona 85721, USA concomitant with progesterone reduced the level of 3a,5a-THP in the central nervous 1. Jorde, L. B., Bamshad, M. & Rogers, A. R. BioEssays 20, 126–136
system from 8.2 6 1.3 ng per g to 2.3 6 2. Majumder, P. P. Evol. Anthropol. 6, 100–110 (1998).
0.35 ng per g without altering levels of 5a- 3. Bamshad et al. Hum. Biol. 68, 1–33 (1996).
DHP or progesterone. This treatment, 4. Tambia, S. J. in The Character of Kinship (ed. Goody, J.) 191–229 which prevented the withdrawal of 3a,5a- (Cambridge Univ. Press, 1973).
5. Mantel, N. Cancer Res. 27, 209–220 (1967).
THP that normally accompanies proges- 6. Heyer, E. et al. Hum. Mol. Genet. 6, 799–803 (1997).
terone withdrawal, completely prevented 7. Saitou, N. & Nei, M. Mol. Biol. Evol. 4, 406–425 (1988).
the withdrawal effects of progesterone we 8. Anderson et al. Nature 290, 457–465 (1981).
9. Wakeley, J. J. Mol. Evol. 37, 613–623 (1993).
observed1, including acceleration in the 10. Hammer, M. F. et al. Mol. Biol. Evol. 15, 427–441 (1998).
decay of GABA-gated current and upregu- lation of the GABA -receptor a4 subunit.
Figure 1 Indomethacin increases stress-induced We have shown8 that direct withdrawal of Effects of progesterone 3a,5a-THP levels in rat cerebral cortex. Male 3a,5a-THP, in the absence of progesterone Sprague–Dawley rats were habituated to handling withdrawal, also results in faster decay of or neuroactive steroid? and injection. Indomethacin (0.1 mg per kg, intraperi- GABA-gated current in association with toneal) was administered 20 minutes before ethanol upregulation of the a4 subunit. In this case, injection stress (2 g per kg, intraperitoneal) and withdrawal from the GABA-modulatory Smith et al.1 reported some interesting 3a,5a-THP levels were measured as described4.
metabolite 3a,5a-THP was accomplished results about the effects of progesterone Indomethacin increased stress-induced 3a,5a-THP with the use of a 5a-reductase inhibitor to treatment and withdrawal on the gene levels by 58.5 6 62.0 % (P , 0.01, Student's t-test).
block 3a,5a-THP formation without alter- encoding the GABA receptor a4 subunit, ing progesterone levels. Taken together, a constituent of receptors for the neuro- numerous steroids other than 3a,5a-THP.
these results establish 3a,5a-THP as the transmitter GABA (g-aminobutyric acid).
Therefore, even if 3a,5a-THP levels are active agent that exhibits withdrawal prop- However, we disagree with their conclusion reduced by indomethacin, this does not erties. In contrast, Morrow et al. administer that the effects of progesterone withdrawal demonstrate that the effects of progesterone indomethacin 20 minutes before ethanol are mediated by reduced levels in the brain withdrawal on GABA receptors are medi- stress, which would prevent direct conver- of the GABA receptor active neurosteroid ated by 3a,5a-THP. Correlational evidence, sion of 3a-DHP to 3a,5a-THP. The although suggestive, does not establish a indomethacin-induced increases in 3a,5a- or allopregnanalone). The ability of THP levels they report after ethanol stress indomethacin to reverse the effects of The effects of progesterone and ethanol can be explained only by stress-induced progesterone was interpreted as evidence withdrawal on GABA receptor a4 subunit effects on the degradation of 3a,5a-THP.
that the effects of progesterone were medi- gene expression are similar. However, We further disagree with the suggestion ated by 3a,5a-THP. Smith et al.1 did not cross-tolerance between ethanol and 3a,5a- by Morrow et al. that our findings are purely provide direct evidence that levels of THP does not occur in vivo. Indeed, chronic correlational. Indomethacin is a highly 3a,5a-THP were reduced by indomethacin ethanol administration produces cross-tol- selective blocker of 3a-hydroxysteroid oxi- (which reversed the effects of proges- erance to benzodiazepines, but also results doreductase activity9, and so reduces 3a,5a- terone), or that progesterone levels were in sensitization to both behavioural and THP levels specifically, without altering lev- not altered by indomethacin. In some neurochemical responses to 3a,5a-THP4.
els of other hormones such as 5a-DHP or cases, indomethacin can increase proges- Moreover, this effect is greater in female progesterone. This is therefore the most terone levels directly2, which may explain than male rats, suggesting that the higher definitive procedure to determine the role of why indomethacin can reverse the effects of progesterone levels in female rats are associ- the 3a,5a-THP metabolite in the upregula- ated with greater sensitization to 3a,5a- tion of the a4 subunit we observe after prog- The effect of indomethacin blockade of THP5. This sensitization is probably due to esterone withdrawal. Direct administration 3a-hydroxysteroid oxidoreductase activity intrinsic changes in GABA receptors, as of 3a,5a-THP would be a less convincing is dependent on the relative levels of 3a,5a- concentrations of 3a,5a-THP are not model because back-conversion to 5a-DHP THP and its immediate precursor 5a- altered after chronic ethanol administra- can occur9. Furthermore, systemic adminis- dihydroprogesterone. When 3a,5a-THP tion4. Although the effects of ethanol4,6 and tration of progesterone best mimics physio- concentrations are raised after progesterone progesterone on GABA receptor a4 gene logical conditions in our animal model of infusion or stress, indomethacin would expression are known, there is no direct evi- premenstrual syndrome, in which circulat- block the oxidation of 3a,5a-THP, increas- dence that 3a,5a-THP alters the expression ing progesterone is converted in the CNS to ing its concentration. Using the same dose of this gene.
regimen as ref. 1 we find that stress-induced A. Leslie Morrow*, Margaret J. VanDoren,
Earlier results from Morrow et al. indi- elevation of 3a,5a-THP levels in the brain Leslie L. Devaud
cate that sensitivity to 3a,5a-THP is are increased by pretreatment with Departments of Psychiatry and *Pharmacology, increased after ethanol withdrawal4. We are indomethacin (Fig. 1). Blockade of this and Center for Alcohol Studies, intrigued by this finding because recent data enzyme with fluoxetine has also been University of North Carolina School of Medicine, from our laboratory indicate that withdraw- shown to raise 3a,5a-THP levels in the rat Chapel Hill, North Carolina 27599-7178, USA al of 3a,5a-THP also increases ethanol brain3. The omission of 3a,5a-THP and potentiation of GABA-gated current in CA1 progesterone levels in the report by Smith et hippocampus by 71.2 6 5.4% over control al.1 therefore raises concerns about the con- Smith et al. reply — Morrow et al. are levels. This result is consistent with reports clusion that 3a,5a-THP mediates the effects incorrect to state that we did not provide of increased alcohol consumption in women of progesterone withdrawal.
direct evidence that 3a-OH-5a-pregnan- during the premenstrual period. Both of Manipulation of progesterone levels in 20-one (3a,5a-THP, or allopregnanolone) these findings clearly distinguish the effects female rats is likely to influence the expres- levels were reduced by indomethacin. We of ethanol from those of benzodiazepines1,8, sion of many genes as well as the levels of pointed out1 that our previous studies7 which exhibit reduced GABA-modulatory Nature Macmillan Publishers Ltd 1998
NATURE VOL 395 15 OCTOBER 1998 www.nature.com

Source: http://jorde-lab.genetics.utah.edu/elibrary/Bamshad_1998.pdf