Preferential X chromosome loss but random inactivation characterize primary biliary cirrhosis


  • Monica Miozzo,

    1. Medical Genetics Unit, San Paolo Hospital School of Medicine, University of Milan, Milan, Italy
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Carlo Selmi,

    1. Division of Internal Medicine and Liver Unit, San Paolo Hospital School of Medicine, University of Milan, Milan, Italy
    2. Division of Rheumatology, Allergy and Clinical Immunology, University of California at Davis, Davis, CA
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Barbara Gentilin,

    1. Medical Genetics Unit, San Paolo Hospital School of Medicine, University of Milan, Milan, Italy
    Search for more papers by this author
  • Francesca R. Grati,

    1. Medical Genetics Unit, San Paolo Hospital School of Medicine, University of Milan, Milan, Italy
    2. Cytogenetics and Molecular Biology Unit, TOMA Laboratories, Busto Arsizio, Varese, Italy
    Search for more papers by this author
  • Silvia Sirchia,

    1. Medical Genetics Unit, San Paolo Hospital School of Medicine, University of Milan, Milan, Italy
    Search for more papers by this author
  • Sabine Oertelt,

    1. Division of Internal Medicine and Liver Unit, San Paolo Hospital School of Medicine, University of Milan, Milan, Italy
    2. Division of Rheumatology, Allergy and Clinical Immunology, University of California at Davis, Davis, CA
    Search for more papers by this author
  • Massimo Zuin,

    1. Division of Internal Medicine and Liver Unit, San Paolo Hospital School of Medicine, University of Milan, Milan, Italy
    Search for more papers by this author
  • M. Eric Gershwin,

    1. Division of Rheumatology, Allergy and Clinical Immunology, University of California at Davis, Davis, CA
    Search for more papers by this author
  • Mauro Podda,

    1. Division of Internal Medicine and Liver Unit, San Paolo Hospital School of Medicine, University of Milan, Milan, Italy
    Search for more papers by this author
  • Pietro Invernizzi

    Corresponding author
    1. Division of Internal Medicine and Liver Unit, San Paolo Hospital School of Medicine, University of Milan, Milan, Italy
    • Division of Internal Medicine and Liver Unit, San Paolo Hospital School of Medicine, University of Milan, via di Rundiní 8, 20142, Milan, Italy
    Search for more papers by this author
    • fax: (39) 0250323089

  • Potential conflict of interest: Nothing to report


Recent work has demonstrated enhanced X monosomy in women with primary biliary cirrhosis (PBC) as well as two other female-predominant autoimmune diseases, systemic sclerosis and autoimmune thyroid disease. To further our understanding of these events, we have investigated the mechanisms of X chromosome loss and X chromosome inactivation (XCI) in 166 women with PBC and 226 rigorously age-matched healthy and liver disease controls. X chromosome analysis and determination of loss pattern was performed by quantitative fluorescent polymerase chain reaction (QF-PCR) with 4 X-linked short tandem repeats. Further definition of the XCI was based on analysis of methylation-sensitive restriction sites. Importantly, in PBC the X chromosome loss occurs not only more frequently but also in a preferential fashion. This observation supports our thesis that the enhanced X monosomy involves only one parentally derived chromosome and is not secondary to a constitutive non random pattern of XCI. In fact, in the presence of monosomy, the lost X chromosome is necessarily the inactive homologue. Conclusion: The finding that the X chromosome loss is preferential suggests the critical involvement of X chromosome gene products in the female predisposition to PBC and also emphasizes the need to determine the parental origin of the maintained chromosome to investigate the role of imprinting. (HEPATOLOGY 2007.)

Similar to the majority of autoimmune conditions, primary biliary cirrhosis (PBC) is characterized by a striking female predominance.1 Importantly, the critical importance of genetic susceptibility to PBC is supported by the high concordance rate in monozygotic twins,2 the tendency to observe multiple cases in pedigrees,3 and recent observations in case reports of specific mutations.4 We do know, however, that PBC is likely a combination of genetic and environmental factors.5, 6 Indeed, we previously submitted that sex chromosome defects would provide clues in the search for loci involved in PBC susceptibility and sex ratios7 and have reported that women with PBC and two other female-predominant autoimmune diseases, systemic sclerosis and autoimmune thyroid disease, are characterized by an enhanced X monosomy rate in peripheral blood mononuclear cells (PBMC), particularly T and B lymphocytes8, 9 compared to healthy age-matched women.

The X chromosome shares unique epigenetic characteristics, in that one X homologue is inactivated in women to achieve equivalent levels of X-linked gene products in both sexes. The choice of the X chromosome to be inactivated is normally random and females are thus functional mosaics for X-linked genes, with approximately 50% of cells having one parentally derived X chromosome being active. Non random (e.g., skewed) X chromosome inactivation (XCI), that is the preferential choice of the chromosome to be silenced, can be secondary to mutations or, alternatively, the result of a stochastic event during X inactivation commitment.10 Further, the XCI status is dynamic during life and older females may manifest a skewed pattern as an acquired trait that is neither uncommon nor pathogenic per se.11, 12 Finally, approximately 15% of X-linked genes can escape XCI and display a constitutive biallelic expression in a variable proportion of healthy women.13 Such characteristic genomic differences should be accounted in explaining sex-specific phenotypes in complex diseases.

We hypothesize that an enhanced X monosomy in PBMC may cause not only haploinsufficiency of X-linked genes escaping XCI, but also specific hemizygosity of one of the two X chromosomes in women with PBC. This hypothesis can be supported by two possible pathways: a non random X chromosome loss, that is the preferential involvement of one of the two parentally-derived X homologues (Fig. 1, panel B), or a skewed XCI pattern (Fig. 1, panel A). It is important to emphasize that in presence of a X loss event, the X homologue destined to be lost is necessarily the inactive one, in order to maintain an adequate X-linked gene dosage for cell survival; because of this, it is important to determine whether or not an observed X preferential loss event is secondary to a skewed XCI status.

Figure 1.

Suggested mechanisms of X chromosome inactivation (XCI) and X loss patterns in women are illustrated. The two parentally derived X chromosome homologues are depicted in different colors (blue, yellow); inactive chromosomes are included in grey boxes. (A) Random and skewed XCI patterns reflect the expected consequences on the subsequent cell generations. (B) Random and preferential X loss patterns lead to generations of cell populations showing both or only one X-linked haplotype, respectively. (C) The hypothesized scenario of women with PBC compared to healthy controls is shown. While PBMC manifest a random XCI pattern, this can be associated with a random (upper part, most common in healthy women) or preferential (lower part, most common in women with PBC) X chromosome loss. These mechanisms lead to the establishment of different immune competent cell lineages characterized by monosomy of the same X chromosome uniquely in women with PBC.


PBC, primary biliary cirrhosis; XCI, X chromosome inactivation; PBMC, peripheral blood mononuclear cells; STR, short tandem repeat.

Materials and Methods


PBMC were isolated from fresh whole blood samples obtained from 166 women with PBC followed at the Liver outpatient clinic of the San Paolo Hospital, University of Milan, Italy. In all cases, the diagnosis of PBC was based on the presence of at least two out of three internationally accepted criteria (i.e., compatible liver histology, positive test for serum antimitochondrial antibodies (AMA) ≥1:40, and/or plasma alkaline phosphatase levels at least one and a half times the upper normal limit for longer than 6 months).1 AMA negative patients otherwise fulfilling such diagnostic criteria14 were included in the study in order to produce a more representative cohort of the PBC population. One hundred thirty-one/166 (79%) patients with PBC were currently taking ursodeoxycholic acid as the sole treatment at the time of enrollment. Table 1 shows the major clinical and biochemical features of the patients with PBC enrolled in this study.

Table 1. Characteristics of Patients with PBC
 All Patients (n =166)
  • NOTE. Continuous variables are expressed as median (range).

  • *

    Complications including ascites, portal-systemic encephalopathy, and gastrointestinal bleeding were observed only in patients with advanced histological stages.

Female sex166/166 (100%)
Age (years)59 (29–84)
Disease duration (years)12 (0–26)
AMA negative14/166 (8%)
Serum albumin (g/dL)4.1 (2.4–4.9)
Total bilirubin (mg/dL)0.9 (0.3–15.4)
No. in histological stages III-IV76/166 (46%)
No. with major portal hypertension complications*14/76 (18%)
No. in ursodeoxycholic acid therapy131/166 (79%)

PBMC samples from 177 healthy women (mean ± standard deviation age 58 ± 13 years), randomly selected from a pool of normal volunteers with no sign or history of autoimmunity or chronic liver disease and 49 women with chronic hepatitis C (CHC) (mean ± standard deviation age 54 ± 16 years) naïve to anti-viral treatments were used as controls. Since age is known to be correlated with both X monosomy rate15 and XCI patterns,11 subjects were subdivided into two groups (≥55 years or younger) for matching purposes to both CHC and healthy women control groups. In addition, women with CHC reproduced the distribution of the three main stages of liver disease severity (i.e., absence of cirrhosis, compensated cirrhosis, and cirrhosis with major complications of portal hypertension) observed in the PBC cohort. Prior to this study, all subjects had been investigated for X monosomy rates in PBMC by fluorescent in situ hybridization (data not shown) as previously described.8

The study protocol followed the ethical guidelines of the most recent Declaration of Helsinki (Edinburgh, 2000) and subjects enrolled in the study provided written informed consent.

X Chromosome Analyses

DNA was extracted from PBMC from all enrolled women using the phenol-chloroform-isoamylic alcohol method.16 We used 4 X-linked short tandem repeats (STRs) (HUMARA, DXS8105, DXS996, P39) to determine the X chromosome loss pattern by quantitative PCR and 2 STRs for XCI pattern analysis (HUMARA and DXS6673E) using methylation-sensitive enzymes.

X Chromosome Loss Pattern.

The X loss pattern was investigated as X chromosome dosage by Quantitative Fluorescent PCR (QF-PCR) analysis, a validated method applied for the rapid diagnosis of prenatal aneuploidy,17 and herein used to assess allelic imbalance in a subset of cells. The HUMARA locus, located at the androgen receptor gene (AR; Xq12), was genotyped in PBMC from 166 PBC women and from 177 healthy women. To rule out the possibility of a partial chromosome loss we evaluated three additional STRs (DXS8105, DXS996 and P39) spanning both arms of the X chromosome (Fig. 2) in a subgroup of 25 women with PBC and in 35 age-matched controls characterized by high X monosomy rates (≥6.5%). The criteria for the choice of these subgroups were based on the evidence that quantitative fluorescent polymerase chain reaction (QF-PCR) is characterized by low specificity when the aneuploidy is found in a minor subset of cells (i.e., low level mosaicism).15 As we have previously demonstrated,18 the sensitivity of QF-PCR largely depends on the proportion of cells carrying chromosome imbalances and drops significantly when such proportion is lower than 5%. We then reasoned to arbitrarily select cases showing X monosomy rates higher than 6.5% to maximize the probability of detecting X monosomy at molecular level.

Figure 2.

Mapping of the 4 polymorphic loci used to ascertain the X chromosome preferential loss. Allelic profiles from peripheral blood cells of a representative PBC case are depicted on the right.

We have previously reported that the enhanced X monosomy is unique to PBC and not found in CHC.4 To reprove this point, a group of women with chronic viral disease was included in this further analysis to further rule out the possibility that chronic inflammation or liver injury per se interferes with preferential X chromosome loss.

Primer sequences and PCR conditions of HUMARA and DXS8105 (Xp22.33) STRs were obtained from a previous report19 and from the Human Genome Database (, respectively. Conditions for the study of DXS996 (Xp22.31) and P39 (Xq28) STRs followed the QF-PCR protocol reported by Ogilvie and colleagues.20 All amplification products were run on capillary electrophoresis on ABI Prism 310 Sequencer and analyzed, using GeneScan software (both from Applied Biosystems, Foster City, CA). Allelic dosage ratios were obtained as previously described21, 22; the normal ranges for HUMARA and DXS8105 (0.9–1.335 and 1.1–1.45, respectively) were obtained from 155 control women; data for the DXS996 and P39 STR were previously reported.20 Based on QF-PCR results, allelic ratios external to these ranges indicate an imbalanced dosage for one X chromosome, i.e. a preferential loss of one X homologue. Conversely, in the case of random loss, the ratio between the specific X alleles at a polymorphic locus remains balanced.

X Chromosome Inactivation Assay.

The definition of the XCI pattern was based on analysis of methylation-sensitive restriction sites flanked by the specific primers for the HUMARA and DXS6673E STR that allow one to quantitatively identify the inactive (methylated) X chromosome. The HUMARA assay19 was performed in all women DNA samples: i.e., PBC (n = 166), healthy (n = 177) and CHC (n = 49). The analysis of DXS6673E23 was performed only in the cases that were uninformative (i.e., homozygous) at the HUMARA locus. PCR was performed before and after enzymatic digestions of the methylation sensitive restriction sites, using the HpaII and HhaI enzymes (Boehringer Ingelheim, Mannheim, Germany) for HUMARA and HhaI and RsaI for DXS6673E as previously reported.19, 23, 24 All samples were tested in duplicate and one male DNA sample was included in each experiment as a control for enzymatic digestion. PCR products were run by capillary electrophoresis and XCI values determined in heterozygous cases using the formula previously reported.25 This formula calculates the XCI pattern as (d1/u1)/(d1/u1+d2/u2), where d1 and d2 represent the two peak areas from the digested sample and u1 and u2 are the corresponding areas of the alleles obtain from undigested DNA. This method overcomes the risk of imbalanced allelic dosage in DNA not cleaved with methylation sensitive endonucleases. The XCI pattern was defined as “moderately” skewed in the presence of an XCI ratio ≥75:25 and “severely” when the XCI pattern was equal to or exceeded 90:10, similar to previous reports.11, 24

Statistical Analysis

The chi-squared test was used for categorical variables; in the case of 2×2 tables with small expected frequencies (<5), the Fisher's exact test was applied. P values of less than 0.05 were considered as statistically significant and all analyses were two-sided and performed using Stata Statistical Software (Stata Corporation, College Station, TX).


Preferential X Chromosome Loss

HUMARA QF-PCR was informative (heterozygous) in 140/166 PBC, 40/49 CHC cases and 155/177 controls. An imbalanced allelic dosage (i.e., a preferential X loss) was demonstrated in 54/140 (39%) PBC, in 7/40 (17.5%) CHC cases and in 37/155 (24%) controls (P = 0.006) (Table 2). Although imbalance at HUMARA locus suggests the loss of an entire X chromosome, we provided further support to this finding by studying 3 additional STRs (Fig. 2) in a subgroup of 25 patients and 36 healthy controls showing the highest monosomy rates within each group; this analysis increased the power to demonstrate a molecular X chromosome dosage imbalance that was only found in a cell subpopulation. The two subject subsets used in this analysis were homogeneously distributed respect to the age. As summarized in Table 3, for each of the 4, an allelic unbalancing was found more frequently in women with PBC. Furthermore, assuming the presence of at least 3 imbalanced STRs to detect preferential loss of one whole X chromosome, we note that this condition was present in 8/21 PBC cases and in none of the healthy women (P = 0.001). Analysis of PBC subgroups based on disease stage, serum AMA status, and ongoing ursodeoxycholic acid treatment failed to identify an association between these clinical features and preferential X loss pattern (data not shown), thus suggesting that the data are not secondary to subsequent pathology.

Table 2. QF-PCR Results Using HUMARA STR Evaluated on Heterozygous PBC Patients and Controls, Including Healthy Women and Women with Chronic Hepatitis C (CHC)
 Normal RatioImbalanced RatioP Value (vs. PBC)
PBC86/140 (61%)54/140 (39%)
Healthy118/155 (76%)37/155 (24%)0.006
CHC33/40 (82.5%)7/40 (17.5%)NS
Table 3. X Chromosome Imbalance in PBC and Controls Showing X Monosomy ≥6.5%
STRSamples (Number of Informative Cases)Imbalanced Ratio() (%)P Value
  • Abbreviation: NS, not significant.

  • Outlier values with respect to ranges defined in 155 healthy women or previously reported.

  • Only P values < 0.200 are shown.

HumaraPBC (22)14/22 (64)0.002
 Controls (31)6/31 (19) 
DXS8105PBC (21)9/21 (43)NS
 Controls (26)8/26 (31) 
DXS996PBC (19)12/19 (63)0.061
 Controls (23)7/23 (30) 
P39PBC (20)10/20 (50)0.012
 Controls (32)5/32 (16) 
Any STRPBC (25)21/25 (84)0.025
 Controls (35)19/35 (54) 
At least 3 STRPBC (21)8/21 (38)0.001
 Controls (27)0/27 

Random XCI Pattern

As expected, the frequency of non-random XCI increased with age in all groups (Table 4) and the prevalence rates for skewed (mild or severe) XCI in PBMC did not differ significantly between PBC cases and either of the control populations (CHC, healthy).

Table 4. X Chromosome Inactivation (XCI) Patterns
 AgeXCI ≥75% (%)XCI ≥90% (%)
All ages< 55 Years≥ 55 YearsAll Ages< 55 Years≥ 55 Years
PBC59 ± 1270/166 (42%)20/58 (35%)50/108 (46%)26/166 (16%)3/58 (5%)23/108 (21%)
Hepatitis C58 ± 1323/49 (47%)8/18 (44%)15/31 (48%)5/49 (10%)1/18 (6%)4/31 (13%)
Healthy54 ± 1672/177 (41%)29/81 (36%)43/96 (45%)26/177 (15%)10/81 (12%)16/96 (17%)

Co-existence of Skewed XCI and Preferential X Loss

In case of X monosomy the chromosome destined to be lost must be the inactive one. Thus, we investigated whether a preferential X loss was due to the presence of a skewed XCI status in the same subject. Our findings support that the presence of a preferential X chromosome loss is independent of the XCI pattern and only a small subset of patients and controls had both a preferential X loss and a skewed XCI pattern (2.7% of PBC versus 1.7% of control subjects, P = NS).


Several genes known to be crucial for the maintenance of immune tolerance are located on the X chromosome and mutations of specific X-linked genes are well known to be related to several immunodeficiency syndromes.26 In addition, monosomy or major structural abnormalities of the X chromosome, as seen in Turner's syndrome27, 28 and premature ovarian failure,29 are often characterized by autoimmune features. Following our recent findings of an enhanced monosomy X in peripheral lymphocytes of women with PBC or other autoimmune diseases8, 9 and reports of a frequent skewed XCI in women with non hepatic autoimmune diseases,30–32 we have now investigated the mechanisms of X loss and XCI in 166 women with PBC, 49 with CHC and 177 healthy controls. We speculated that a specific hemizygous status of the monosomic X might unmask peculiar haplotypes responsible for susceptibility to autoimmunity. We demonstrate herein that in PBMC from women with PBC the X chromosome loss occurs not only more frequently than in controls, but also in a preferential fashion (Fig. 1, panel C). These findings were first achieved by QF-PCR of the HUMARA locus in 140 informative PBC, 40 CHC subjects and 155 controls and further confirmed by the study of 3 additional X-linked loci in a nested group of subjects showing high monosomy levels. This latter investigation demonstrated imbalance in at least three loci in about 40% of PBC cases characterized by a high monosomy rate but none of the controls, thus demonstrating that loss specifically involves one whole X chromosome. Although we optimized the quantitative PCR experiments to detect chromosome imbalance that is present in a minor subset of peripheral blood cells, preferential X chromosome loss was not observed in all PBC cases, possibly due to the sensitivity limits of the molecular test. We note that these limitations are shared by all molecular techniques aimed at pinpointing abnormalities that are not homogeneously found in DNA samples.

To determine whether a skewed XCI could be an alternative/additional phenomenon also resulting in hemizygosity of X-linked loci we investigated the XCI status in our cohort of women with PBC and controls. The study revealed a similar age-related frequency of skewed XCI in both cases and controls, thus implying that PBC is not associated with an enhanced rate of skewed XCI in contrast with the data in autoimmune thyroid disease and scleroderma.30–32 However, these latter findings were based on limited numbers of informative subjects and did not include age-matched controls; in fact, although variable prevalence rates of a skewed XCI status have been reported, studies taking into account age as a modifier factor, clearly indicate that as many as 16% of healthy women over the age of 50 are characterized by a severe XCI skewing.11, 12, 25, 33 Importantly, we also ruled out the possibility that preferential X loss and skewed XCI could be related to each other. The analysis of the distribution of XCI patterns according to the presence of an X chromosome preferential loss failed to identify any correlation.

Our findings cumulatively support that enhanced X monosomy in PBMC preferentially involves only one parentally-derived chromosome and is not secondary to a constitutive non random pattern of XCI in PBC (Fig. 1, panel C). We therefore submit a model (Fig. 1, panel C) in which the X loss observed in a subset of PBMC from women with PBC is preferential and affects cells otherwise characterized by a homogeneous XCI pattern.

Our data on the mosaic condition of the preferential X loss in PBC are in agreement with proposed models for sex chromosome abnormalities in which genetic defects are acquired rather than constitutively present. Accordingly, we hypothesize that the presence of a peculiar X-linked haplotype, unmasked by X monosomy, could induce autoimmunity in a predisposed subject. Therefore, we submit that hemizygosity of X chromosome loci is a critical point for future genetic studies.34 In fact, a similar approach should be designed to pinpoint specific X-linked haplotypes, most likely among those involving genes variably escaping XCI.

It is important to note that the proposed haploinsufficiency of a subset of X-linked genes could be encountered also in males with PBC, following the loss of the Y chromosome. Indeed, future studies of PBC onset in men should include scrutinizing Y chromosome loci having homologues on the X chromosome. Further steps may be hypothesized and summarized in two different scenarios. If such Y and X chromosome genes escape inactivation, PBC might result in men from a Y chromosome loss resulting in half-dosage of the putative genes. In this regard, candidate genes with these characteristics might be sought within the PAR1 region that includes the short arm telomeres of both sex chromosomes. It will be important to investigate whether sex chromosomes instability is present also in male patients with PBC, that is whether they lack the Y chromosome more frequently compared to healthy men. It is interesting to note that, similarly to what described for the X chromosome, an age-dependent loss of the Y chromosome has also been reported in healthy men.35 Lastly, future studies could focus on determining the paternal origin of the maintained X chromosome to investigate the possible role of imprinting loci on the X chromosome, as already supposed for cognitive function in Turner syndrome.36 Alternatively, haplo-insufficiency of a candidate gene can be present also without Y chromosome homologue if this gene escapes inactivation.