Results

Generation of Xpd Compound Heterozygotes
We generated an Xpd knock-in allele with a point mutation encoding a single amino acid change (XPDG602D) found in the XPCS patient XPCS2 (Figure 1A–1C). mRNA expression from the targeted allele could be detected in embryonic stem cells by RT-PCR (Figure 1D), although expression was reduced approximately 5-fold relative to wt mRNA transcript levels as determined by Northern blotting of RNA from the testis of heterozygous animals (Figure 1E). Because patient XPCS2 was a hemizygote with mutant XPD protein (XPDG602D) expressed from a single allele, the corresponding mutation was expected to be viable in the homozygous state. However, homozygous mutant mice were not observed, neither amongst live births nor embryonic day 13.5 (E13.5) or E3.5 embryos (Table 1). The corresponding hypomorphic, mutant allele was thus designated as homozygous lethal (†XPCS). Homozygous lethality of the XPCS allele is likely due to reduced levels of expression of this essential protein as a result of gene targeting (Figure 1A) rather than to the mutation itself. Xpd ablation (XpdKO /KO ) is similarly incompatible with life beyond the earliest stages of embryogenesis [22]. Consistent with this interpretation, a different targeted Xpd mutation encoding XPDR683W, which is associated with XP in the homozygous state in humans, was similarly underexpressed and lethal in the homozygous state (designated as †XP allele) (Figure 1A–1C; Table 1; unpublished data). Also, a different targeting approach leading to the use of the native 3′UTR and removal of the neo gene resulted in normalisation of XpdXPCS mRNA levels and viable homozygous XpdXPCS/XPCS (XPDG602D/G602D) animals [23].
Figure 1  Targeting of the Mouse Xpd Gene
(A) Schematic representation of the genomic structure and partial restriction map of the wt and targeted mouse Xpd loci. For the wt Xpd allele, shaded boxes represent coding regions of exons 12 and 19–23; the 3′UTR is represented by an open box. TGA indicates the translational stop codon; PolyA indicates the polyadenylation signal. For the XpdTTD targeted allele, the 194–base pair (bp) human XPD cDNA fragment fused to exon 22 is indicated as a striped box including the TTD (R722W) mutation indicated by a vertical arrow. Chicken β-globin exons 2 and 3 including the 3′UTR are indicated as black boxes with corresponding Roman numerals followed by the β-globin polyadenylation signal (PolyA*). For the Xpd†XP and Xpd†XPCS targeted alleles, vertical arrows indicate XPCS (G602D-encoding) and XP (R683W-encoding) mutations in exons 19 and 22, respectively. The unique 3′ probe located outside the targeting construct is marked by a thick black line. Restriction sites: B, BamHI; C, ClaI; E, EcoRI; H, HindIII; Hp, HpaI; Sf, SfiI.
(B) Southern blot analysis of EcoRI-digested genomic DNA from wt, Xpd†XPCS/wt, and Xpd†XP/wt recombinant embryonic stem cell clones hybridised with the 3′ probe depicted in (A). The wt allele yields a 6.5-kilobase (kb) fragment, whereas both targeted Xpd†XP and Xpd†XPCS alleles yield a 5.1-kb fragment.
(C) Genotyping of wt and targeted alleles by PCR using primers F2, R1, and mR as indicated in (A) yields fragments of 399 bp and 468 bp, respectively.
(D) RT-PCR detection of mRNA expression originating from the targeted †XP and †XPCS alleles in embryonic stem cell clones using primers F1 (hybridising outside the targeting construct) and mR as indicated in (A) results in a 1,416-bp fragment.
(E) Northern blot analysis of total RNA isolated from testis of homozygous wt and XpdTTD/TTD, heterozygous Xpd†XPCS/wt and XpdTTD/wt, and compound heterozygous Xpd†XPCS/TTD mice as indicated. Hybridisation with a 1.4-kb mouse Xpd cDNA probe detects mRNAs of 4, 3.3, and 2.7 kb from wt, Xpd†XPCS, and XpdTTD alleles, respectively. An ethidium bromide (EtBr)–stained gel showing the amount of total RNA loaded is shown below.
Table 1  Frequency of Xpd†XP/†XP, Xpd†XPCS/†XPCS, and Compound Heterozygous Xpd†XP/†XPCS Embryos and Offspring

“Null” Allele Can Alleviate Developmental Delay, Skin, and Hair Features of TTD
To test the potential of a homozygous lethal “null” allele to nevertheless contribute to organismal phenotype, we combined an Xpd†XPCS allele with a viable XpdTTD allele by crossing the corresponding heterozygous animals. Similar to hemizygous TTD mice carrying one true Xpd knockout allele (XpdTTD/KO), compound heterozygous XpdTTD/†XPCS mice were born at the expected Mendelian frequencies. Expression from the Xpd†XPCS allele was also reduced in the testis of compound heterozygous animals, whereas expression from the XpdTTD allele was increased relative to wt by ~5-fold (Figure 1E). Because of a lack of available antibodies and the inability to distinguish amongst various mutant forms of XPD differing only by single amino acid substitutions, we were unable to ascertain the relative amount of XPD protein from the different alleles.
Despite reduced levels of mRNA expression, the homozygous lethal Xpd†XPCS allele ameliorated multiple XpdTTD-associated disease symptoms in compound heterozygous XpdTTD/†XPCS animals including the hallmark brittle hair and cutaneous features fully penetrant in homo- and hemizygous TTD mice (Figure 2A–2C). In marked contrast to XpdTTD/TTD (and XpdTTD/KO) mice, which display complete hair loss in the first hair cycle and partial hair loss in subsequent cycles throughout their lives [21], compound heterozygous XpdTTD/†XPCS mice displayed some hair loss only during the first hair cycle and only locally at the back (Figure 2A). Scanning electron microscope analysis of XpdTTD/†XPCS hair revealed an almost normal appearance, with TTD-like features such as broken hairs found only at very low frequency (unpublished data). Amino acid analysis confirmed that cysteine levels in the hair of the XpdTTD/†XPCS mice were significantly higher than in XpdTTD/TTD animals, but remained below the wt level (Figure 2C). TTD hemizygotes (XpdTTD/KO) do not display significant differences in cutaneous features and longevity relative to homozygous XpdTTD/TTD mice [21].
Figure 2  Partial Rescue of TTD Cutaneous, Blood, and Developmental Phenotypes in Compound Heterozygous XpdTTD/†XPCS Mice
(A) Photographs of 5-mo-old homozygous XpdTTD/TTD, compound heterozygous XpdTTD/†XPCS, and wt mice. Insets: images of first-round hair loss.
(B) Histological analysis of the skin of XpdTTD/TTD, XpdTTD/†XPCS, and wt mice. TTD-associated acanthosis (thicker epidermis, indicated by solid vertical line), pronounced granular layer (indicated by arrows), and sebacious gland hyperplasia (indicated by dotted vertical line) were absent in the epidermis of XpdTTD/†XPCS and wt mice. Magnification 400×.
(C) Cysteine content of hair from wt, XpdTTD/TTD, and XpdTTD/†XPCS mice. The p-value indicates significant differences between mutants and wt, as well as between XpdTTD/TTD and XpdTTD/†XPCS mice. Error bars indicate standard error of the mean (SEM).
(D) Hematocrit values from blood of XpdTTD/TTD and XpdTTD/†XPCS mice. The p-values indicate the significance of the difference relative to wt. Error bars indicate SEM.
(E) Body weights of developing XpdTTD/TTD and XpdTTD/†XPCS mice after weaning plotted as a percentage of the weight of age-matched control wt and heterozygote (hz) littermates (set at 100%). Error bars indicate SEM. Other prominent TTD features in the epidermis, including acanthosis (thickening of the layer of the nucleated cells), hyperkeratosis (prominent thickening of the cornified layer), and pronounced granular layer and sebacious gland hyperplasia (causing greasy appearance of the hair), were absent in the skin of XpdTTD/†XPCS mice, as established by blind microscopic examination of skin sections (Figure 2B). Furthermore, anaemia and developmental delay present in patients with TTD [24] and in XpdTTD/TTD mice [15] were both partially rescued in compound heterozygous XpdTTD/†XPCS mice (Figure 2D and 2E).

Rescue of Progeroid Features in TTD Mice by Homozygous Lethal Xpd Alleles
Because patients with TTD, XPCS, and CS (but not XP) and the corresponding mouse models share similar accelerated progeroid symptoms [12,13,15,23], we next addressed ageing-related parameters in compound heterozygous mice (Figure 3). Whereas XpdTTD/TTD animals show reduced bone mineral density as an indication of the early onset of osteoporosis before ~14 mo of age [15], tail vertebrae from compound heterozygous XpdTTD/†XPCS mice were comparable to wt even at 20 mo of age (Figure 3B and 3C). Furthermore, whereas XpdTTD/TTD mice developed kyphosis earlier than wt animals (onset ~3 mo versus 12–20 mo), compound heterozygous XpdTTD/†XPCS mice did not (Figure 3B). Overall appearance and body weight curves revealed that TTD-associated age-related premature cachexia and lack of general fitness were fully rescued in compound heterozygous XpdTTD/†XPCS mice (Figure 3A and 3D). Finally, the life span of compound heterozygotes was extended relative to XpdTTD/TTD mice (Table 2).
Figure 3  Rescue of TTD-Associated Segmental Progeroid Features in Compound Heterozygous Xpd TTD/†XPCS Mice
(A) Photographs of 20-mo-old wt, compound heterozygous XpdTTD/†XPCS, and homozygous XpdTTD/TTD mice. Note the extreme cachexia (lack of subcutaneous fat) in the XpdTTD/TTD mouse and the absence of this phenotype in wt and XpdTTD/†XPCS mice.
(B) Radiographs of 20-mo-old male wt, XpdTTD/†XPCS, and XpdTTD/TTD mice. Ageing XpdTTD/TTD mice develop kyphosis (curvature of the spinal column) and reduction of bone mineral density as shown in the 6–8 segment of the tail vertebrae counted from the pelvis (see close-up at right). Note the absence of these features in the XpdTTD / † XPCS mouse.
(C) Quantification of relative bone mineral density of tail vertebrae from 20-mo-old male wt (n = 3), XpdTTD/†XPCS (n = 4), and XpdTTD/TTD (n = 3) mice. The p-values indicate the significance of the difference relative to XpdTTD/TTD. Error bars indicate SEM.
(D) Body weight curves as a function of time. Note that the age-dependent cachexia observed in XpdTTD/TTD mice was rescued in both male and female XpdTTD / †XPCS mice. Significant differences between wt and XpdTTD/TTD but not between wt and XpdTTD/†XPCS mice were observed at 9 and 18 mo of age as indicated by asterisks. Error bars indicate SEM.
Table 2  Pleiotropic Xpd Biallelic Effects in Mice and Cells To determine whether the homozygous lethal Xpd†XPCS allele was unique in its ability to ameliorate symptoms associated with the XpdTTD allele, we generated compound heterozygous XpdTTD/†XP mice by crossing the corresponding heterozygous animals. Similar to the Xpd †XPCS allele, the homozygous lethal Xpd †XP allele rescued cutaneous symptoms including hair loss (except locally during the first round; unpublished data), reduced cysteine content (cysteine index 9.3 ± 0.9 standard deviation [87% of wt], p = 0.01 versus TTD), ageing-associated premature cachexia (males and females were 36.1 ± 6.4 g [93% of wt] and 39.2 ± 3.2 g [116% of wt], respectively), and reduced life span (Table 2). Taken together, these data indicate that two independent alleles, which on their own are unable to support viability (Table 1), were nonetheless able to ameliorate TTD-associated phenotypes in vivo (Table 2).

Molecular Mechanisms of Biallelic Effects
We next turned to UV-based cellular assays including unscheduled DNA synthesis after UV irradiation (UV-UDS), recovery of RNA synthesis after UV irradiation (UV-RRS), and UV survival, which report on the NER subpathways (global genome NER and transcription-coupled NER) and total NER, respectively. In none of these assays was the response to UV improved in compound heterozygotes relative to TTD homozygotes (Figure 4A–4C). However, unlike the in vivo TTD phenotypes described above, in which XpdTTD/TTD and XpdTTD/KO animals were indistinguishable, XpdTTD dosage effects were observed in UV survival, UV-UDS, and UV-RRS, indicating that cellular parameters as measured in fibroblasts here do not always correlate with the phenotype at the level of the intact organism. XpdTTD/KO hemizygous cells were thus used as the baseline on which to compare the activity of compound heterozygous cells. Relative to XpdTTD/KO hemizygote cells, UV survival was improved by the homozygous lethal Xpd†XPCS allele in XpdTTD/†XPCS compound heterozygous cells and to a lesser degree by the Xpd†XP allele (Figure 4A). Because of embryonic and cellular lethality, we were unable to test UV survival associated exclusively with the Xpd†XPCS or Xpd†XP alleles. However, homozygous XPDXP (XPDR683W) and hemizygous XPDXPCS (XPDG602D) human cells are known to be highly sensitive to UV [19,25], as are cells from a homozygous viable XpdXPCS/XPCS (XPDG602D/G602D) mouse model (Figure 4A, dotted line) [23]. Thus, the survival of XpdTTD/†XPCS (and XpdTTD/†XP) cells likely represents a level of UV resistance that neither mutant allele can impart on its own (Table 2). Significant effects of compound heterozygosity on NER subpathways relative to XpdTTD/KO cells were observed in XpdTTD/†XP cells but only for UV-UDS activity. Finally, none of the mutant TFIIH combinations (carrying alterations associated with TTD [XPDR722W], XPCS [XPDG602D], or XP [XPDR683W]) exhibited synergism in an in vitro NER reaction reconstituted with different mutant TFIIH complexes (Figure 4D). Taken together, these data are consistent with interallelic complementation of UV sensitivity in cells but underscore the lack of any correlation between UV-related repair characteristics and TTD progeroid phenotypes in animal models.
Figure 4  TFIIH Functions and Mechanisms of XPD-Associated Disease Pleiotropy
(A) Cellular survival after UV irradiation. Rescue of hemizygous XpdTTD/KO survival by Xpd†XPCS and Xpd†XP alleles is illustrated by arrows marked A and B, respectively. UV survival of homozygous XpdXPCS/XPCS cells (asterisk) from the normally expressed viable allele (XpdXPCS) is depicted by a dotted line. Survival curves represent an average of four independent experiments; 1–2 cell lines per genotype were included in each experiment. Error bars indicate SEM between experiments.
(B) UV-UDS, a measure of global genome repair. Number of experiments: n = 15 (XpdTTD/TTD), n = 6 (XpdTTD/KO), n = 4 (XpdTTD/†XPCS ), n = 2 (XpdTTD/†XP ); 1–2 cell lines per genotype were included in each experiment. The asterisk indicates significant difference with XpdTTD/TTD; crosses indicate significant differences with XpdTTD/KO.
(C) UV-RRS, a measure of transcription-coupled repair of UV-induced lesions. Number of experiments: n = 7 (XpdTTD/TTD), n = 2 (XpdTTD/KO), n = 4 (XpdTTD/†XPCS ), n = 2 (XpdTTD/†XP ); 1–2 cell lines per genotype were included in each experiment.
(D) Incision/excision activity of combinations of altered TFIIH complexes in a reconstituted NER reaction. Equal amounts of single or mixed populations of recombinant TFIIHs (containing XPD, XPB, p62, p52, His-p44, Flag-p34, cdk7, cyclin H, Mat1, and p8) were mixed with recombinant XPG, XPF/ERCC1, XPC/hHR23B, RPA, and a radiolabelled synthetic NER substrate. The excision products (26–34 nucleotides in length) were visualised at nucleotide resolution on a denaturing polyacrylamide gel as indicated . Note the weak activity corresponding to each single and combined TFIIH complex (lanes 3–8) relative to the wt (lane 1) and negative controls (lane 2).
(E) Xpd dose-dependent reduction of TFIIH in homozygous XpdTTD/TTD, hemizygous XpdTTD/KO, and compound heterozygous XpdTTD/†XPCS and XpdTTD/†XP cells by comparative immunofluorescence of the p62 subunit of TFIIH. Roman numerals represent different microscopic slides and Arabic numerals different cell lines labelled as follows: (I) wt cells (1) labelled with 2-μm beads, XpdTTD/TTD cells (2) with 0.79-μm beads, and XpdTTD/KO cells (3) with no beads; (II) wt cells (1) labelled with 0.79-μm beads and XpdTTD/†XPCS cells (4) with no beads; and (III) wt cells (1) labelled with 0.79-μm beads and XpdTTD/†XP cells (5) with no beads.
(F) Quantification of immunofluorescent signal from at least 50 nuclei per cell line and 2–6 experiments per genotype. Bars representing cells analysed on the same microscopic slide are depicted side by side, with wt set at 100%. The p-value indicates minimum significant difference between wt and the indicated cell lines analysed on the same microscopic slide within one experiment. Next we asked whether the Xpd†XPCS and Xpd†XP alleles, despite decreased mRNA levels, ameliorated TTD symptoms by increasing overall TFIIH levels in compound heterozygous XpdTTD/ †XPCS and XpdTTD/ †XP cells. Previously, using comparative immunohistochemistry, we and others have shown an up to 70% reduction of TFIIH levels in primary fibroblasts from patients with TTD compared with wt controls due to reduced stability [16,17]. Despite overexpression of mRNA from the XpdTTD allele relative to the wt allele (Figure 1E), TFIIH protein levels were reduced by 50% in primary mouse XpdTTD/TTD fibroblasts (Figure 4E and 4F), thereby mimicking the situation in human patients with TTD. In accordance with the gene dosage, a further reduction of up to 70% of the wt level was observed in hemizygous XpdTTD/KO cells. Consistent with low mRNA expression levels, neither the Xpd†XPCS nor the Xpd†XP allele was able to restore TFIIH abundance to wt levels in XpdTTD compound heterozygote cells (Figure 4E and 4F). Thus, the improved UV survival observed in compound heterozygote cells (Figure 4A) and likely the rescue of TTD progeroid symptoms (Figure 3) were not due to normalisation of TFIIH levels, suggesting a qualitative rather than a quantitative effect on these phenotypes in vivo.
In contrast, the level of XPCS mRNA expression did affect the ability of the encoded protein (XPDG602D) to restore the TTD hair phenotype to normal. Notably, XpdTTD/ †XPCS animals had a partial TTD hair phenotype, correlating with low levels of Xpd†XPCS expression, whereas XpdTTD/XPCS animals had wt hair, correlating with normal expression levels from the viable XpdXPCS allele (Table 2 and unpublished data). Thus, the range of expression levels from these two mutant alleles affected their ability to complement some phenotypes (hair). An overview of the functional relationships between Xpd alleles, phenotypes, and the presumed underlying TFIIH function in mice and cells is presented in Table 2.