REVIEW X-Inactivation Analysis in the 1990S: Promise and Potential

REVIEW

X-inactivation analysis in the 1990s: promise and potential problems L Busque1 and DG Gilliland2 ˆ ´ ´ ´ 1Research Center, Hematology Division, Hopital Maisonneuve-Rosemont, Universite de Montreal, Montreal, Canada; and 2Howard Hughes Medical Institute, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA

An important recent contribution to the field of X- Keywords: clonality; X-inactivation; skewing; HUMARA; hematopo- inactivation analysis has been the identification by Allen et iesis al7 of differential methylation sites between the active (Xa) and inactive X (Xi) chromosome closely linked to the multiallelic CAG short tandem repeat8 (STR) of the first exon of the human Introduction androgen-receptor gene (AR or HUMARA; GenBank M21748 JO4150). The AR CAG repeat is polymorphic in 90% of X-inactivation clonality analysis is a clever application of females of all racial groups9 rendering X-inactivation analysis Mary Lyon’s fundamental observation that each X-chromo- feasible in the majority of females with a single assay. The some in excess of one is randomly inactivated in cells of the HUMARA assay has already been validated by several investi- 1 developing female embryos. The first clonality studies perfor- gators10–19 and found to be reliable and reproducible to study med to investigate the nature of human neoplasms were done human neoplasia. The high informativeness of this PCR-based 2–4 almost three decades ago, and have opened a fertile field assay coupled with the ability to precisely quantitate allelic of investigation that has allowed investigators to unravel some ratios have allowed for the analysis of rare diseases such as of the basic tenets of modern oncology. The current decade Langerhans’ cell histiocytoses20 (LCH) and juvenile myelo- has been particularly prolific with the largest number of clon- monocytic leukemia21 (JMML), and the analysis of a large ality studies performed yet, and the accumulation of an unpre- cohort of normal females to investigate the incidence of non- cedented quantity of data. This can be primarily attributed to random X-inactivation.22 Furthermore, it has been possible to the development of newer, more informative clonality assays, analyze X-inactivation pattern in families to investigate herita- which allow X-inactivation analysis to be performed in the bility of skewed patterns of X-inactivation.23 The HUMARA majority of females. Although the new information has assay is currently the most frequently utilized method of X- sometimes confirmed and sometimes contradicted earlier inactivation analysis by most investigators. conclusions, it has mainly opened many novel avenues of Another important improvement has been the development investigation in cancer biology and hematopoiesis. of several clonality assays based on transcriptional analysis of In this paper, we will review the recent progress made in coding polymorphisms. Prchal et al24,25 have been instrumen- X-inactivation techniques, in the etiology of nonrandom X- tal in developing such assays. One of these new assays takes inactivation occurring in the normal population, and in the advantage of a more frequent polymorphisms at the G6PD study of human neoplasms. We will outline the questions and locus (C/T1311) which can be analyzed by reverse potential problems that should be addressed in the next transcription followed by ligase detection reaction. A second decade. informative coding polymorphism has been identified in the palmitoylated membrane protein p55 (G/T at cDNA 358). When both assays are used, heterozygosity of close to 50% The development of newer, more informative X- is expected.26 Another coding polymorphism located in the inactivation-based clonality assays iduronate-2-sulfatase (IDS) gene (C/T 438)27 has also proven to be useful in the study of clonal hematopoietic disorders.28 The applicability of X-inactivation analysis has been mainly Although these assays are substantially less informative than limited by intrinsic properties of the methods used to deter- the HUMARA assay, they have the theoretical advantage of mine X-inactivation ratios in females. The electrophoretic not relying on methylation pattern to assess the activity of the analysis of the glucose-6-phosphate dehydrogenase isoen- X-chromosome, since they directly rely on the exclusive tran- 5 zymes (G6PD) used by Fialkow and coworkers was limited scriptional activity of the active X-chromosome. by a low level of informativeness in the general population, which effectively precluded analysis of rare disorders, analysis of large cohorts of females, and familial analysis. In the mid Study of the normal female population: unequal 1980s, informativeness was improved by the identification of Lyonization revisited DNA polymorphisms linked to residues differentially methyl- 6 ated between the active and inactive X-chromosome. The Incidence of nonrandom X-inactivation (NRXI) in the informativeness was increased but still insufficient to allow normal population general applicability. Unequal Lyonization, which is most appropriately termed ˆ nonrandom X-inactivation (NRXI), is said to occur when there Correspondence: L Busque, Research Center, Hopital Maisonneuve- ´ ´ Rosemont, 5415, de l’Assomption Blvd, Montreal, Quebec, Canada, is a significant deviation from the theoretical 1:1 ratio between H1T 2M4; Fax: 514 254 – 5094 the paternal (Xp) and maternal (Xm) X-chromosome in healthy Received 12 November 1997; accepted 20 November 1997 females. Although criteria for NRXI are arbitrary, an allelic Review L Busque and DG Gilliland 129 ratio у3:1 (which corresponds to the expression of 75% of one allele) has been widely accepted in the literature.13,22,29–32 Determination of the incidence of NRXI should be straightfor- ward since it can be simply estimated by using X- inactivation assays in healthy females. Surprisingly, data obtained in the last 20 years by different investigators have been contradic- tory. Vogelstein et al32 found only three out of 81 (3.7%) normal females to have NRXI using the phosphoglycerate kinase (PGK) and hypoxanthine phosphoribosyl transferase (HPRT) X- inactivation probes. Recently, Gale et al29 found significant NRXI in blood-derived cells in 23% of normal females using PGK and HPRT probes and 22% of females using M27␤,30 which has been confirmed in numerous other studies.33–36 Discordance in the incidence of NRXI was initially explained by the diversity of assays used, the different criteria for NRXI, and the small population sizes. However, a more plausible explanation came from studies that analyzed X- Figure 1 Incidence of nonrandom X-inactivation (NRXI) in blood inactivation patterns of different tissues in the same female. cells (allelic ratio у3:1) of normal females from three different age For example, Fey et al33 reported that the incidence of NRXI groups. The incidence of NRXI is 37.9% in females aged more than 60 years and statistically different from the 30 years (P Ͻ 0.0064) and was low in gastrointestinal mucosa and thyroid tissue, but was from the neonate group (P Ͻ 0.0001) (from Busque L et al22). significantly higher in blood cells. Gale et al31 reported that 45% of females analyzed have different patterns of X-inactivation when blood-derived cells were compared to muscle and skin samples. Interestingly, blood cells in both studies rently, NRXI occurring in normal females is thought to be were more frequently skewed than other tissues. These studies related to the limited number of cells present at the time of suggested that X-inactivation patterns were tissue-specific. random X-inactivation during embryogenesis.5,29 In this One explanation for specificity of the X-inactivation patterns model, the variance of the distribution of X-inactivation pat- was elucidated by the direct analysis of X-inactivation using terns in females follows a binomial distribution where the the X-linked lacZ transgene, which is subjected to inactivation deviation from the mean is inversely proportional to the num- and consequent loss of ␤-galactosidase activity in mouse ber of cells present at the time of X-inactivation. For example, embryos. These studies showed that X-inactivation proceeds if the inactivation occurs when there are only four cells with different schedules in different somatic tissues.37 present, the incidence of skewing (ratio у3:1) will be 37%. If The second, and perhaps most important clue to the dis- inactivation occurs at the 16 cell stage, the incidence of skew- crepancy in incidence of NRXI came with the analysis of X- ing will be 5%.43 In contrast to humans, NRXI has been shown inactivation patterns in females of different age groups. An age to be genetically controlled in murine species. Cattanach et difference in NRXI was first suggested by Fey et al38 who al44 demonstrated that the different alleles of a gene called reported a higher incidence of skewing in older females com- Xce (X-chromosome controlling element) can bias the choice pared to children. We characterized the effect of age on NRXI of the X to be inactivated. Xce has three well characterized in a large cross-sectional study in which the X-inactivation alleles which can be classified as a gradient of strength: Xcec ratios were determined with the HUMARA assay in 295 nor- Ͼ Xceb Ͼ Xcea.45 The heterozygotes exhibit NRXI. Xce maps mal females from three age groups: (1) neonates; (2) 28–32 within or in the vicinity of the Xic (X-chromosome inactivation years old; and (3) age greater than 60 years. We documented center),46 and it has been shown biochemically45 and cytolog- an incidence of skewing (ratio у3:1) of 8.6% in neonates, ically47 to cause primary NRXI rather than secondary cell 16.4% in the 28 to 32 years and 37.9% in women greater selection. Furthermore, the effects of different Xce alleles can than 60 years (P Ͻ 0.0001 vs neonates, P = 0.064 vs 28 to modify the imprinted preferential inactivation of Xp in extra- 32 years) (Figure 1).22,39 The different incidence of skewing embryonic tissue.48 between the neonates and the 30-year-old group was not stat- The investigation of heritability of NRXI in humans has been istically significant. This suggests that skewing found in nor- precluded by the low informativeness of most assays. The mal neonates probably correspond to true NRXI occurring multiple alleles of the STR at the AR locus has allowed for during the embryological process of X-inactivation, and that the evaluation of the transmission of skewed patterns of X- skewing found in older females is likely to be caused by other inactivation in families. Naumova et al23 have reported a fam- mechanisms (vide infra). These results have subsequently been ily where all seven daughters had highly skewed patterns of confirmed by other investigators.40–42 X-inactivation with preferential activation of the paternal X- Since the incidence of skewing may vary more than four- chromosome. Furthermore, the paternal grandmother also fold depending on the age of the population, it is clear that showed a skewed pattern of X-inactivation with the same X- the age of the cohort analyzed may constitute the major chromosome preferentially activated. These results suggested confounding factor explaining the discrepancy of previous an X-linked transmission of NRXI in this family. Pegoraro et studies. al49 reported a high incidence of NRXI in a family and localized the genetic anomaly responsible for the trait to the Xq28 region using linkage analysis. They further demonstrated The etiology of nonrandom X-inactivation (NRXI) in that affected individuals have a large segment of DNA deleted the normal population in the Xq28 region.49 Further evidence for a genetic cause of NRXI came with the recent work by Plenge et al50 who Primary nonrandom X-inactivation: The cause of NRXI documented a rare cytosine to guanine mutation in the XIST has been more difficult to investigate than its incidence. Cur- (X-inactivation specific transcript) minimal promoter in nine Review L Busque and DG Gilliland 130 females of two different families that all showed preferential clonally derived. The hypothesis of a human X-linked gene inactivation of the X chromosome carrying the mutation. affecting the growth of hematopoietic cells was first suggested Although the data cited above clearly indicate that NRXI by Luzzato et al.54 may be genetically determined and transmissible in some fam- Convincing recent evidence for such a genetically deter- ilies, it does not rule out a stochastic cause of NRXI for the mined positive growth selection advantage has been provided vast majority of females. To evaluate the possibility of herita- by Abkowitz et al55 through X-inactivation analysis at the bility of NRXI in the general population of females, we have G6PD locus of Safari cats (F1 generation of Goeffroy (G) and analyzed the relationship between the degree of NRXI in 177 Domestic (d) cats). Balanced hematopoiesis was documented mother–daughter (neonates) pairs, of which 142 were both between the G and d alleles in the first year of life of the heterozygous at the AR locus. The allelic ratio between the heterozygous cats, but most of these animals demonstrated two X-linked alleles at the AR locus was determined for each skewed patterns of X-inactivation in blood-derived cells with individual. Allelic ratios for the neonatal population were aging. These results are identical to the acquired skewing scored into two categories: skewed (ratio у3, n = 17) or non- phenomenon found in humans. The intrinsic qualities of the skewed (ratio Ͻ3, n = 125). Of the 17 skewed neonates, 10 model developed by Abkowitz et al has made it possible to (58.8%) had mothers that were also skewed. In contrast, of distinguish between a stochastic and a genetically controlled the 125 non-skewed neonates, only 26 (20.8%) had mothers cause of skewing. Since no recombination is possible between that were skewed (P = 0.0018, Fisher’s exact test, two-tail) (R the two Xs of the Goeffroy and Domestic cats, the Safari cats Mio and L Busque, 1997 submitted). have half their cells with the G X-chromosome active and half These data, taken together, support the hypothesis that with the d X-chromosome active. If the cause of the skewing NRXI may be due not only to the stochastic process of X- phenomenon is stochastic (clonal hematopoiesis or stem cell inactivation occurring in a limited number of cells, but that depletion) there is as much chance for the phenomenon to one or several genes may bias the choice of which X is to be occur in G or in d cells, and consequently there should be as inactivated during the embryological process of X- many cats skewed in favor of the G allele as cats skewed for inactivation. Further studies should be aimed at identifying the the d allele. All the cats studied by Abkowitz et al showed a genetic elements responsible for NRXI. predominant G allele, which provides persuasive evidence that the acquired skewing is genetically determined, rather than stochastic. Furthermore, these data indicate that the Acquired skewing in blood cells with aging: The increased skewing gene must be localized to the X-chromosome, other- incidence of nonrandom X-inactivation ratios found in blood wise no skewing could be observed. The fact that in the initial cells of the female elderly population, termed acquired skew- phase of hematopoiesis, X-inactivation was balanced between ing, may be caused by a completely different mechanism than the G and the d alleles also eliminates primary NRXI NRXI documented in neonates. Several hypotheses may (embryological) as a cause of skewing. The mechanism would explain the acquired skewing phenomenon. First, it is possible be the opposite of the one described for skewing that occurs that acquired skewing truly corresponds to clonal hematopo- in females heterozygous for X-linked disease genes (reviewed iesis caused by somatic mutation(s) in hematopoietic stem in Ref. 56). In these rare females, the cells that harbor the cell(s) conferring a growth advantage to the mutated cell(s). mutant X in the active state are selected against. An argument in favor of this possibility is that the incidence of hematological malignancies increases significantly with age. Age-related incidence is particularly striking for diseases Study of human neoplasia such as myelodysplastic syndrome (MDS) and acute myelogenous leukemia (AML) with a peak incidence after 65 years The availability of more informative tests for clonality assess- of age.51–53 However, the incidence of 40% of skewing in the ment such as the HUMARA assay has allowed for analysis normal population in this age group would argue against any of numerous cancers including hematologic malignancies and pathological association, since even the most common hema- several solid tumors with benign and malignant behavior. It tologic malignancies have incidences less than 0.01%. has also been possible to analyze reactive disorders, nononco- Another hypothesis is that acquired skewing is a reflection of logic conditions such as nodular hyperplasia of liver,57 and stochastic clonal dominance secondary to significant stem cell even atherosclerosis58 (Table 1). We will provide a brief over- depletion with age in normal females. If this hypothesis is view of several recent findings. applied to the acquired skewing in normal females observed The distinction between polyclonality and clonality is fun- in our study, it would suggest that 40% of the normal females damental to the pathophysiology of any disease process. Clon- have a severe reduction in the number of their stem cells. A ality infers selective growth advantage caused by acquired third possibility is that the skewing phenomenon is a genetic somatic mutation occurring in a single progenitor cell. Field trait caused by a gene which resides on one of the two X- effect, viral infection, stimulation by cytokines may cause chromosomes. In contrast to the gene involved in primary polyclonal growth of cells. Documentation of clonality may NRXI, this gene would not bias the choice of which X-chro- orient research toward the identification of specific mutations mosome is to be inactivated, but instead would provide a causing the disease. X-inactivation analysis has recently selective growth advantage to hematopoietic cells that have allowed for the elucidation of the clonal nature of several this gene on the active X-chromosome. If the growth advan- enigmatic disorders. One such disease is Langerhans’ cell his- tage is slight, it may take years for the population of cells with tiocytoses (LCH) which was considered by many as a noncan- this gene in the active state to completely dominate the mar- cerous reactive disorder.59 LCH had eluded clonality investi- row. In this circumstance, all cells would have the same X in gation in the past because of its relative rarity, the absence of the active state and will show a completely skewed pattern of a specific marker allowing to distinguish the lesional histio- X-inactivation compatible with the acquired skewing cytes from reactive cells, and the difficulty of analyzing a pure phenomenon. In this model, the cells would originate from population of cells. The high informativeness of the HUMARA numerous different progenitors, and therefore would be poly- assay coupled with the ability to measure precisely X-inacti- Review L Busque and DG Gilliland 131 Table 1 Recent X-inactivation clonality studies

Field of investigation Tissue or disease Assay used Results or comments Ref.

Clonality assays DNA HUMARA Heterozygosity of 90%, applicable in 7 all nucleated cells RNA HUMARA Limited by level of expression in 10 different tissue RNA G6PD Heterozygosity of 27% 24 RNA p55 Heterozygosity of 48% 25 RNA IDS Heterozygosity 25–30% 27 DNA Fragile X (FMR1) Heterozygosity 45–65% 69 Normal females Blood cells PGK, HPRT High incidence of skewing in the 29 normal population Blood cells, thyroid, gastric PGK, HPRT, M27␤ Tissue speciﬁcity of X-inactivation 33 mucosa patterns PMNs, T cells, colonic mucosa M27␤ Tissue speciﬁcity of X-inactivation 31 patterns Blood cells M27␤ Skewing more frequent in elderly 38 females Blood, three age groups HUMARA, PGK Acquired skewing in elderly females 22 (0/30/Ͼ60) PMNs, T cells of elderly females HUMARA Acquired skewing more frequent in 41 PMNs than in T cells PMNs, T cells different age HUMARA Acquired skewing more frequent in 42 group PMNs than in T cells Hematologic Langerhans’ cells histiocytis HUMARA Clonal CD1a cells 60 malignancies Langerhans’ cells histiocytis HUMARA, PGK, M27␤ Clonal lesional histiocytes in all 20 subsets of disease Juvenile myelomonocytic HUMARA, PGK, G6PD Clonal proliferation of an early 21 leukemia hematopoietic progenitor Essential thrombocythemia (ET) HUMARA, IDS, G6PD, ET is mostly a clonal disorder; some 28 p55 cases polyclonal CD3 negative large granular PGK Clonal derivation of LGL 70 lymphocyte (LGL) Rosai–Dorfman disease HUMARA Polyclonal population of histiocytes 63 Autologous bone marrow HUMARA Clonal hematopoiesis precedes the 68 transplantation development of secondary myelodysplastic syndrome Solid tumors Kaposi’s sarcoma HUMARA Monoclonality and single origin of 64 multicentric lesions Kaposi’s sarcoma HUMARA Polyclonality 71 Meningiomas HUMARA Most clonal, some polyclonal 12 Gangliomas HUMARA Neuronal and glial cells are clonally 66 derived and of common origin Uterine leomyomata HUMARA Clonal; independent origin of 65 individual tumors Medullary carcinoma of thyroid HUMARA Polyclonality in hereditary and 62 sporadic cases Miscellaneous Wiskott–Aldrich syndrome HUMARA Skewing in early lineage of female 72 carrier Focal nodular hyperplasia of HUMARA Clonal development of hyperplasia of 57 liver liver Atherosclerosis HUMARA Clonality of smooth muscle cells 58 Parathyroid tumors in uremia M27␤ Monoclonality in several refractory 73 hyperplasia

This list of studies is not exhaustive and does not include all clonality studies performed in the 1990s, but relates to the topics discussed in the text.

vation ratios has demonstrated that LCH is a clonal proliferat- AR, PGK and G6PD loci.21 Other examples of diseases where ive disorder of CD1a cells. This has been shown directly by monoclonality was documented in situations where polyclon- the analysis of CD1a flow sorted cells60 and indirectly by the ality was expected include: (1) Focal nodular hyperplasia of analysis of tissue infiltrated by lesional CD1a cells.20 Similarly, the liver (FNH), a benign hepatic lesion composed of an juvenile myelomonocytic leukemia (JMML)61 which was encapsulated multinodular proliferation of normal appearing thought possibly to reflect polyclonal proliferation of hemato- hepatocytes, has been shown to result from clonal prolifer- poietic cells with increased sensitivity to GM-CSF, has only ation of hepatocytes using the HUMARA assay.57 (2) It has recently been shown to be the result of clonal proliferation of been documented that the atherosclerotic plaque that occurs early stem cell progenitors by the analysis of clonality at the in coronary arteries is comprised of clonally derived smooth Review L Busque and DG Gilliland 132 muscle cells.58 In the latter situation, it is possible that the Table 2 Causes of nonrandom X-inactivation in normal females unexpected clonality is the consequence of the proliferation (excluding clonal proliferative disorders) of several smooth muscle progenitor cells originating from a small region or patch of the artery where all cells have the Primary (embryological) Stochastic: due to the small number of cells present at the same pattern of X-inactivation. In contrast to the examples time of X-inactivation cited above, it has also been possible to demonstrate that Xq28 deletion some diseases of expected monoclonality have been shown Mutation in the promotor of XIST to be polyclonally derived. One such disease is the medullary XCE? human equivalent of murine Xce carcinoma of the thyroid gland in multiple endocrine neo- Carrier state of X-linked mutant alleles plasia type 2, which has been shown to be polyclonally Agammaglobulinemia derived in the majority of cases including sporadic and heredi- X-linked severe combined immunodeficiency tary forms.62 It is also interesting to note that Rosai–Dorfman’s Wiskott–Aldrich syndrome disease, or histiocytosis with massive lymphadenopathy, X-linked dyskeratose congenita X-linked adrenoleukodystrophy which shows several similarities to LCH has been shown to Incontinentia pigmenti be caused by polyclonal proliferation of cells, implying that X-linked ␣-thalassemia with mental retardation syndrome this disease has a different pathogenesis than LCH.63 (ATRX) X-inactivation analysis also has the unique capacity to Acquired skewing of blood cells in elderly females investigate the cellular origin of multicentric neoplasms. Do all disseminated tumors of a particular cancer originate from a single mutant progenitor cell? Or, do they arise synchron- ously from independent mutated cells? This distinction is fun- cancers based solely on clonal proliferative characteristics. It damental to our understanding of the mechanism of tumor is also important to note that clonality per se should not influ- metastasis, and to the documentation of a potential field effect ence the therapeutic approaches to a disease. For example the in certain cancers. One recent study was aimed at resolving therapeutic approach to Langerhans’ cell histiocytoses59 with this issue for multicentric Kaposi’s sarcoma lesions. The clon- indolent clinical courses, such as eosinophilic granuloma of ality analysis at the AR locus documented that skin lesions bone, should not necessarily be influenced by the recent of Kaposi’s sarcoma were clonally derived.64 It was further documentation that these are clonal proliferative disorders. documented that multiple lesions in the same individual had The relationship between clonality and malignancy may also the same pattern of X-inactivation. These data are consistent be evaluated in clinical settings where prospective evaluation with a single cell origin for Kaposi’s sarcoma, and suggesting of clonality is feasible. Fortunately, clinical settings where that the mutation(s) giving rise to the cellular growth there is a subgroup of normal females with a sufficiently high advantage occurred prior to systemic dissemination of Kapo- risk of developing hematologic cancer to allow a prospective si’s sarcoma.64 In contrast to the situation of Kaposi’s sarcoma, analysis of samples are rare. However, female patients with Marshal et al65 documented that multiple uterine leiomyomata lymphoma treated with autologous bone marrow transplan- were clonally derived but that the different ‘tumorlets’ in an tation (ABMT) may constitute one such high-risk group, and individual did not have the same X-chromosome in the active offer a rare opportunity to study prospectively the develop- state, arguing for an independent origin of each tumor.65 X- ment of clonal hematopoiesis and future malignant transform- inactivation analysis has also been proven useful to investigate ation. It has been recently shown that this cohort of patients the pathogenesis of tumors composed of different cell types. have an incidence of developing a myelodysplastic syndrome Zhu et al66 investigated rare tumors of the central nervous or leukemia as high as 18% within 6 years of transplant.67 system composed of glial and neuronal components called Prospective analysis of blood samples from this patient popu- gangliogliomas. Using both DNA and RNA-based10 clonality lation has demonstrated clonal hematopoiesis precedes and determination at the AR locus, they have shown ganglio- predicts the development of MDS.68 These data also confirm gliomas to be clonally derived and that both cellular the multistep pathogenesis of secondary leukemia. Analysis of components probably originate from a common mutated sequential samples for candidate oncogenes or tumor sup- progenitor.66 pressor genes may allow for identification of genes which are These novel findings on clonality have promoted a re- involved in each step of the progression from clonal hemato- evaluation of our understanding of the relationship between poiesis to malignant transformation. clonality and malignancy. One major contribution of X-inactivation analysis during the past 30 years has been the documentation that almost all cancers are clonal and that clonality The future of X-inactivation analysis is intimately related to malignancy. However, one of the les- sons of the last decade has been that a skewed pattern of X- What is the role of X-inactivation clonality analysis in the next inactivation does not automatically indicate clonal derivation decade? X-inactivation clonality analysis will continue to be of cells, as there are many other causes of skewing in the nor- an important adjuct to the molecular biology of human neo- mal population (Table 2). Furthermore, the documentation of plasia and can be used to address a spectrum of problems, clonal derivation of cells, in the presence of appropriate including polyclonal vs monoclonal origins of cells. Clonality somatic controls, does not necessarily indicate a malignant analysis has evolved in the last decade as a powerful means process. Even if clonality is a first step toward the progression of investigating disease pathophysiology that is no longer lim- to a malignant phenotype, some clonal proliferations of cells ited to the field of oncology. The biology of hematopoiesis, apparently have no transformation potential. Malignancy per- the mechanism of X-inactivation, and the characterization of haps relates more to the biologic behavior of a population of genetic changes in relation to clonality development are new cells in relation to the host, than to the clonal derivation of avenues of investigation that need to be explored by X- cells. It is clear that several clonal proliferative disorders are inactivation. quite indolent and should perhaps not be considered as true There are several important issues that should be addressed Review L Busque and DG Gilliland 133 by X-inactivation analysis in the next decade. (1) Clonality References assays: extensive validation of the third generation assays to 1 Lyon MF. Sex chromatin and gene action in the mammalian X- precisely delineate their limits and pitfalls. Large cohorts of chromosome. Am J Hum Genet 1962; 14: 135–148. normal and females with known clonal disorders should be 2 Linder D, Gartler SM. Glucose-6-phosphate dehydrogenase analyzed with HUMARA and transcriptional assay such as mosaicism: utilization as a cell marker in the study of leiomyomas. the G6PD, p55 and IDS. (2) Nonrandom X-inactivation Science 1965; 150: 67–69. (embryological): the potential existence of a human equivalent 3 Beutler E, Collins Z, Irwin LE. Value of genetic variants of glucose- to the murine Xce locus that would be responsible for NRXI 6-phosphate dehydrogenase in tracing the origin of malignant tumors. New Engl J Med 1967; 276: 389–391. should be investigated by linkage analysis in families. The 4 Fialkow PJ, Gartler SM, Yoshida A. Clonal origin of chronic myelo- identification of alleles of different strength that bias the cytic leukemia in man. Proc Natl Acad Sci USA 1967; 58: choice of which X is to be inactivated during embryogenesis 1468–1471. would allow, for the first time, the distinction between NRXI 5 Fialkow PJ. Primordial cell pool size and lineage relationships of from clonal derivation of cells in the absence of any specific five human cell types. Ann Hum Genet 1973; 37: 39–48. somatic control tissue. (3) Acquired skewing in blood cells of 6 Vogelstein B, Fearon ER, Hamilton SR, Fienberg AP. Use of restric- tion fragment length polymorphism to determine the clonal origin elderly females: the cause of acquired skewing in normal of human tumors. Science 1985; 227: 642–645. females should be investigated since this may give an 7 Allen RC, Zoghbi HY, Moseley AB, Rosenblatt HM, Belmont JW. important insight on the biology of the aging hematopoietic Methylation of HpaII and HhaI sites near the polymorphic CAG tissue and possibly in the biology of age-associated hemato- repeat in the human androgen-receptor gene correlates with X logical malignancy. The acquired skewing phenomenon may chromosome inactivation. Am J Hum Genet 1992; 51: 1229– provide unique insights into the biology of hematopoiesis in 1239. 8 La Spada AR, Wilson EM, Lubahn DB, Harding AE, Fischbeck KH. humans. (4) Study of human neoplasia: X-inactivation assays Androgen receptor gene mutation in X-linked spinal and bulbar should be used as molecular markers of disease progression muscular atrophy. Nature 1991; 352: 77–79. and help define the genetic events responsible for acquisition 9 Edwards A, Hammond HA, Jin L, Casey T, Chakraborty R. Genetic of a cellular growth advantage, and which other genetic alter- variation at five trimeric and tetrameric repeat loci in four human ations are responsible for the transformation toward a full population groups. Genomics 1992; 12: 241–253. malignant phenotype. X-inactivation analysis is essential to 10 Busque L, Zhu J, DeHart D, Griffith B, Willman C, Carroll R, Black PMcL, Gilliland DG. An expression based clonality assay at the the elucidation of the multistep pathogenesis of cancer. human androgen receptor locus (HUMARA) on a chromosome X. Finally, application of clonality analysis to disease processes Nucleic Acids Res 1994; 22: 697–698. thought to be polyclonal such as artherogenesis, may yield 11 Zhu J, Leon PL, Beggs AH, Busque L, Gilliland DG, Black PMcI. further surprises and insights into pathophysiology of these Human pituitary adenomas show no loss of heterozygosity at the diseases. (5) Diagnostic applications of X-inactivation clonal- retinoblastoma gene locus. J Clin Endocrinol Metab 1994; 78: ity assays: the recent accumulation of data should allow us to 922–927. 12 Zhu, J, Frosh PM, Busque L, Beggs AH, Dasher K, Gilliland DG, devise clinical application of such assays in the near future. Black PM. Analysis of meningiomas by methylation-based and The possibility of diagnosing a clonal population of cells in transcription-based clonality assay. Cancer Res 1995; 55: 3865– any tissue in most females would be a tremendous break- 3872. through for woman’s health. The extensive validation of 13 Allen RC, Nachtman RG, Rosenblatt HM, Belmont JW. Appli- assays, a better understanding of the causes of skewing in nor- cation of carrier testing to genetic counseling for X-linked agam- mal females and in depth understanding of the biology of maglobulinemia. Am J Hum Genet 1994; 54: 25–35. 14 Mansfield ES, Blasband A, Kronick MN, Wrabetz L, Kaplan P, Rap- clonal proliferative disorders are prerequisites to the applica- paport E, Satore M, Parrella T, Surrey S, Fortina P. Fluorescent bility of X-inactivation assays in the clinical setting. approaches to diagnosis of Lesch–Nyhan syndrome and quantitat- ive analysis of carrier status. Mol Cell Probes 1993; 7: 311–324. 15 Anan K, Ito M, Misawa M, Ohe Y, Kai S, Kohsaki M, Hara H. Conclusions Clonal analysis of peripheral blood and haematopoietic colonies in patients with aplastic anaemia and refractory anaemia using the polymorphic short tandem repeat on the human androgen recep- It has been nearly three decades since Fialkow harnessed an tor (HUMARA) gene. Br J Haematol 1995; 89: 838–844. extraordinary source of insight into the pathophysiology of 16 Mutter GL, Boynton KA. PCR bias in amplification of androgen human neoplasia through analysis of X-chromosome inacti- receptor alleles, a trinucleotide repeat marker used in clonality vation. Only rarely does one individual open a field of scien- studies. Nucleic Acids Res 1995; 23: 1411–1418. tific inquiry that provides so many seminal insights into such 17 Guerrasio A, Rosso C, Martinelli G, Lo Coco F, Pampinella M, a broad spectrum of human disease processes. His work has Santoro A, Lanza C, Allione B, Resegotti L, Saglio G. Polyclonal hematopoiesis associated with long-term persistence of the AML1- opened fertile new venues for investigation, and will remain ETO transcript in patients with the FAB M2 acute myeloid leuke- the foundation for our understanding of the clonal prolifer- mia in continuous clinical remission. Br J Haematol 1995; 90: ation nature of human cancers. Dr Fialkow has provided the 364–368. shoulders for us to stand on. In his memory, we will hope to 18 El Kassar N, Hetet G, Li Y, Briere J, Grandchamp B. Clonal analysis catch a glimpse of the horizon. of hematopoietic cells in essential thrombocythemia. Br J Haema- tol 1995; 90: 131–137. 19 Gale RE, Mein CA, Linch DC. Quantification of X-chromosome inactivation patterns in haematological samples using the DNA Acknowledgement PCR-based HUMARA assay. Leukemia 1996; 10: 362–367. 20 Willman CH, Busque L, Griffith B, Favara BE, McLain KL, Duncan ´ LB is a scholar of the Fonds de la Recherche en Sante du MH, Gilliland DG. Langerhans cell histiocytosis (histiocytose X): a ´ Quebec (FRSQ). DGG is a Stephen Birnbaum Scholar of the clonal proliferative disease. New Engl J Med 1994; 331: 154–160. Leukemia Society of America and is an Assistant Investigator 21 Busque L, Gilliland DG, Perchal JT, Sieff CA, Weinstein HJ, Sokol JM, Belichova M, Wayne AS, Zukerman KS, Sokol L, Castleberry of the Howard Hughes Medical Institute. The authors would RP, Emanuel PD. Clonality in juvenile chronic myelogenous leu- like to thank Sylvie Provost MSc for assistance in preparing kemia. Blood 1995; 85: 21–30. this manuscript. 22 Busque L, Mio R, Mattioli J, Brias E, Blais N, Lalonde Y, Maragh Review L Busque and DG Gilliland 134 M, Gilliland DG. Nonrandom X-inactivation patterns in normal 43 Gale RE, Linch DC, Interpretation of X-chromosome inactivation females: lyonization ratios vary with age. Blood 1996; 88: 59–65. patterns. Blood 1994; 84: 2376 (letter). 23 Naumova AK, Plenge RM, Bird LM, Leppert M, Morgan K, Willard 44 Cattanach BM. Control of X chromosome inactivation. Ann Rev HF, Sapienza C. Heritability of X chromosome-inactivation pheno- Genet 1975; 9: 1–18. type in a large family. Am J Hum Genet 1996; 58: 1111–1119. 45 Johnston PG, Cattanach BM. Controlling elements in the mouse 24 Prchal JT, Guan YL. A novel clonality assay based on transcrip- IV. Evidence of non-random X-inactivation. Genet Res 1981; 37: tional analysis of the active X chromosome. Stem Cells 1993; 11: 151–160. 62–65. 46 Cattanach BM, Rasberry C, Evans EP, Dandolo L, Simmler M-C, 25 Luhovy M, Liu Y, Belickova M, Prchal JF, Prchal JT. A novel clon- Avner P. Genetic and molecular evidence of an X-chromosome ality assay based on transcriptional polymorphism of X chromo- deletion spanning the tabby (Ta) and testicular feminization (Tfm) some gene p55. Bio Blood Marrow Transplant 1995; 1: 81–87. loci in the mouse. Cytogenet Cell Genet 1991; 56: 137–143. 26 Liu Y, Phelan J, Go RCP, Prchal JF, Prchal JT. Rapid detection of 47 Rastan S. Primary nonrandom X-inactivation caused by controlling clonality by detection of two closely linked X-chromosome exonic elements in the mouse demonstrated at the cellular level. Genet polymorphisms using allele-specific PCR. J Clin Invest 1997; 99: Res 1982; 40: 139–147. 1984–1990. 48 Rastan S, Cattanach BM. Interaction between the Xce locus and 27 Bunge S, Steglich C, Zuther C, Beck M, Morris CP, Schwinger E, imprinting of the paternal X chromosome in mouse yolk-sac endo- Schinzel A, Hopwood JJ, Gal A. Iduronate-2-sulfatase gene derm. Nature 1983; 303: 635–637. mutations in 16 patients with mucopolysaccharidosis type II 49 Pegoraro E, Whitaker J, Mowery-Rushton P, Surti U, Lanasa M, (Hunter syndrome). Hum Mol Genet 1993; 2: 1871–1875. Hoffman EP. Familial skewed X inactivation: a molecular trait ` 28 El-Kassar N, Hetet G, Briere J, Grandchamps B. Clonality analysis associated with high spontaneous abortion rate maps to Xq28. Am of hematopoiesis in essential thrombocythemia: advantage of J Hum Genet 1997; 61: 160–170. studying T lymphocytes and platelets. Blood 1997; 89: 128–134. 50 Plenge RM, Hendrich BD, Schwartz C, Arena JF, Naumova A, Sap- 29 Gale RE, Wheadon H, Linch DC. X-chromosome inactivation pat- ienza C, Winter RM, Willard HF. A promoter mutation in the XIST terns using HPRT and PGK polymorphisms in hematologically gene in two unrelated families with skewed X-chromosome inactivation. Nature Genet 1997; 17: 353–356. normal and post-chemotherapy females. Br J Haematol 1991; 79: ` 193–197. 51 Oscier D. Myelodysplastic syndromes. Ballieres Clin Haematol 30 Gale RE, Wheadon H, Linch DC. Assessment of X-inactivation pat- 1987; 1: 389–426. terns using the hypervariable probe M27␤ in normal hematopo- 52 Sandler DP. Epidemiology of acute myelogenous leukemia. Semin ietic cells and in acute myeloid leukemic blasts. Leukemia 1992; Oncol 1987; 14: 359–364. 6: 649–655. 53 Spangrude GJ, Smith L, Uchida N, Ikuta K, Heimfeld S, Friedman 31 Gale RE, Wheadon H, Boulos P, Linch DC. Tissue specificity of J, Weissman IL. Mouse hematopoietic stem cells. Blood 1991; 78: X-chromosome inactivation patterns. Blood 1994; 83: 2899–2905. 1395–1402. 32 Vogelstein B, Fearon ER, Hamilton SR, Preisinger AC, Willard HF, 54 Luzzatto L, Usanga EA, Bienzle U, Esan GFJ, Fasuan FA. Imbal- Michelson AM, Riggs AD, Orkin SH. Clonal analysis using recom- ance in X-chromosome expression: evidence for a human X-linked binant DNA probes for the X-chromosome. Cancer Res 1987; 47: gene affecting growth of hematopoietic cells. Science 1979; 205: 4806–4813. 1418–1420. 33 Fey MF, Peter H-J, Hinds HL, Zimmermann A, Liechti-Gallati S, 55 Abkowitz JL, Persik MT, Shelton GH, Catlin SN, Guttorp, P, Kikle- Gerber H, Suder H, Tobler A. Clonal analysis of human tumors vich JV. Evidence that an X chromosome gene (closely linked to with M27␤, a highly informative polymorphic X chromosomal G6PD) controls hematopoietic stem cell (HSC) behavior. Blood probe. J Clin Invest 1992; 89: 1438–1444. 1996; 88: 445a (Abstr. 1768). 34 Brown RM, Brown GK. X-chromosome inactivation and the diag- 56 Belmont JW. Insights into lymphocyte development from X-linked immune deficiencies. Trends Genet 1995; 11: 112–116. nosis of X-linked disease in females. J Med Genet 1993; 30: 57 Gaffey MJ, Lezzoni JC, Weiss LM. Clonal analysis of focal nodular 177–184. hyperplasia of the liver. Am J Pathol 1996; 148: 1089–1096. 35 Harris A, Collins J, Vetrie D, Cole C, Bobrow M. X-inactivation 58 Murry CE, Gipaya CT, Bartosek T, Benditt EP, Schwartz SM. as a mechanism of selection against lethal alleles: further investi- Monoclonality of smooth muscle cells in human atherosclerosis. gation of incontinentia pigmenti and X-linked lymphoproliferative Am J Pathol 1997; 151: 697–706. disease. J Med Genet 1992; 29: 608–614. 59 Lipton JM. Histiocytic syndromes. In: Hoffman R, Benz EJ, Shattil 36 Puck JM, Steward CC, Nussbaum RL. Maximum-likelihood analy- SJ, Furie B, Cohen HJ (eds). Hematology Basic Principles and Prac- sis of human T cells X chromosome inactivation patterns: normal tice, first edn. Churchill Livingstone: New York, 1991, pp 604– women vs carriers of X-linked severe combined immunodefi- 615. ciency. Am J Hum Genet 1992; 50: 742–748. 60 Yu RC, Chu C, Buluwela L, Chu AC. Clonal proliferation of Lang- 37 Tan SS, Williams EA, Tarm PPL. X-chromosome inactivation erhans cells in Langerhans cell histiocytosis. Lancet 1994; 343: occurs at different times in different tissues of the post-implan- 767–768. tation mouse embryo. Nat Genet 1993; 3: 170–174. 61 Emanuel PD, Shannon KM, Castelberry RP. Juvenile myelomono- 38 Fey MF, Liechti-Gallati S, von Rohr A, Borisch B, Theilkas L, cytic leukemia: molecular understanding and prospects for ther- Schneider V, Oestreicher M, Nagel S, Ziemiecki A, Tobler A. apy. Mol Med Today 1996; 2: 468–475. Clonality and X-inactivation patterns in hematopoietic cell popu- 62 Ferraris AM, Mangerini R, Gaetani GF, Romei C, Pinchera A, Pac- lation detected by the highly informative M27b DNA probe. Blood ini F. Polyclonal origin of medullary carcinoma of the thyroid in 1994; 83: 931–938. multiple endocrine neoplasia type 2. Hum Genet 1997; 99: 39 Busque L, Brais E, Mattioli J, Blais N, D’Angelo G, Maragh M, 202–205. Gilliland DG. Incidence of excessive lyonisation increase with age 63 Paulli M, Bergamaschi G, Tonon L, Viglio A, Rosso R, Facchetti in hematopoietic cells in normal females. Blood 1994; 84: 368a F, Geerts ML, Magrini U, Cazzola M. Evidence for a polyclonal (Abstr. 1457). nature of the cell infiltrate in sinus histiocytosis with massive lym- 40 Fielding AK, Gale RE, Linch DC. Analysis of X chromosome inacti- phadenopathy (Rosai–Dorfman disease). Br J Haemotol 1995; 91: vation patterns in the elderly; implications for stem cell usage. 415–418. Blood 1995; 86: 590a (Abstr. 2347). 64 Rabkin CS, Janz S, Lash A, Coleman AE, Musaba E, Liotta L, Biggar 41 Champion KM, Gilbert JGR, Asimakopoulos FA, Hinshelwood S, RJ, Zhuang Z. Monoclonal origin of multicentric Kaposi’s sarcoma Green AR. Clonal haemopoiesis in normal elderly women: impli- lesions. New Engl J Med 1997; 336: 988–993. cations for the myeloproliferative disorders and myelodysplastic 65 Mashal RD, Schoenberg Fejzo ML, Friedman AJ, Mitchner N, syndromes. Br J Haematol 1997; 97: 920–926. Nowak RA, Rein MS, Morton CC, Sklar J. Analysis of androgen 42 Gale RE, Fielding A, Harrison CN, Linch DC. Acquired skewing of receptor DNA reveals the independent clonal origins of uterine X-chromosome inactivation patterns in myeloid cells of the elderly leiomyomata and the secondary nature of cytogenetic aberrations suggests stochastic clonal loss with age. Br J Haematol 1997; 98: in the development of leiomyomata. Genes Chromos Cancer 512–519. 1994; 11: 1–6. Review L Busque and DG Gilliland 135 66 Zhu JJ, Leon SP, Folkerth RD, Guo SZ, Wu JK, Black PM. Evidence erts BE, Scott CS. Clonality of CD3 negative large granular lym- for clonal origin of neoplastic neuronal and glial cells in ganglio- phocyte proliferations determined by PCR-based X-inactivation gliomas. Am J Pathol 1997; 151: 565–571. studies. J Clin Pathol 1994; 47: 399–404. ´ ´ 67 Stone R. Myelodysplastic syndrome after autologous transplan- 71 Delabesse E, Oksenhendler E, Lebbe C, Verola O, Varet B, Turhan tation for lymphoma: the price of progress. Blood 1994; 83: AG. Molecular analysis of clonality in Kaposi’s sarcoma. J Clin 3437–3440. Pathol 1997; 50: 664–668. 68 Legare RD, Gribben JG, Maragh M, Hermanowski-Vosatka A, 72 Wengler G, Gorlin JB, Williamson JM, Rosen FS, Bing DH. Non- Roach S, Tantravahi R, Nadler LM, Gilliland DG. Prediction of random inactivation of the X chromosome in early lineage hema- therapy-related acute myelogenous leukemia (AML) and myelo- topoietic cells in carriers of Wiskott–Aldrich syndrome. Blood dysplastic syndrome (MDS) after autologous bone marrow trans- 1995; 85: 2471–2477. plant (ABMT) for lymphoma. Am J Hematol 1997; 56: 45–51. 73 Arnold A, Brown MF, Urena P, Gaz RD, Sarfati E, Drueke TB. 69 Lee ST, McGlennen RC, Litz CE. Clonal determination by fragile X Monoclonality of parathyroid tumors in chronic renal failure and (FMR1) and phosphoglycerate kinase (PGK) gene in hematological in primary parathyroid hyperplasia. J Clin Invest 1995; 95: malignancies. Cancer Res 1994; 54: 5212–5216. 2047–2053. 70 Kelly A, Richards SJ, Sivakumaran M, Shiach C, Stewart AD, Rob-