Conditional ablation of the Mat1 subunit of TFIIH in Schwann cells provides evidence that Mat1 is not required for general transcription

The 36 kDa Mat1 protein (Devault et al., 1995; Fisher et al., 1995; Tassan et al., 1995) along with cyclin H and Cdk7 forms a trimeric kinase complex that has been implicated in the activation of cyclin-dependent kinases (Cdks) during cell cycle progression (reviewed by Kaldis, 1999). The Cdk7-cyclin H-Mat1 trimer has been shown to phosphorylate and activate numerous Cdks in vitro including Cdc2/Cdk1, Cdk2, Cdk3, Cdk4, and Cdk6 (Kaldis, 1999). Genetic experiments in Drosophila melanogaster (Larochelle et al., 1998), fission yeast Schizosaccharomyces pombe (Hermand et al., 2001), and in early mouse embryos (Rossi et al., 2001) support a role for Cdk7-cyclin H-Mat1 in the activation of Cdks and cell cycle progression.

In addition to existing as a trimer in cells, Cdk7, cyclin H, and Mat1 also comprise the kinase subunit of the 9-subunit basal transcription factor IIH [TFIIH (reviewed in Egly, 2001)]. Biochemical purification coupled with recombinant reconstitution experiments have indicated that TFIIH can be purified and function in several subcomplexes; a core TFIIH complex comprised of five subunits (XPB, p62, p52, p44 and p34), and the Cdk7-cyclin H-Mat1 kinase complex (Egly, 2001). The remaining XPD subunit can be found associated with either the core TFIIH or with the kinase subunit and appears to be the means by which the kinase subunit associates with the core to form the TFIIH holoenzyme (Busso et al., 2000; Sandrock and Egly, 2001). The central coiled coil domain of Mat1 has been suggested to mediate the contact between trimeric Cdk7-cyclin H-Mat1 and the XPD helicase which is in turn bridged to the core TFIIH through interaction with the p44 subunit (Busso et al., 2000; Sandrock and Egly, 2001).

TFIIH is known to function in RNA polymerase II (pol II) mediated transcription where, in concert with at least five other basal transcription factors (TFIIB, TFIID, TFIIE and TFIIF), it is critically involved in transcription initiation and promoter escape (reviewed in Dvir et al., 2001). The helicase activity of the XPB subunit has been shown to be critical in both of these functions, and appears to be the only enzymatic activity of TFIIH absolutely required for open complex formation and promoter escape (Moreland et al., 1999; Tirode et al., 1999). Although the XPD helicase is dispensable for RNA synthesis, it nevertheless facilitates optimal transcription activity (Moreland et al., 1999; Tirode et al., 1999). TFIIH also functions in transcription-coupled nucleotide excision repair (NER) where both the XPD and XPB helicases are critical but where the role of the kinase subunit is still in debate (Egly, 2001).

The most widely studied substrate of TFIIH kinase is the C-terminal domain (CTD) of the large subunit of RNA polymerase II. The mammalian CTD consists of 52 repeats of a consensus heptapeptide YSPTSPS, and is efficiently phosphorylated in vitro by TFIIH kinase on serine at position 5 [Ser-5 (Gebara et al., 1997; Sun et al., 1998; Trigon et al., 1998)]. In addition to TFIIH kinase, numerous other kinases phosphorylate the CTD to mediate many aspects of pol II transcription including inhibition of transcription, promoter escape, transcript maturation and pol II processivity (reviewed in Oelgeschlager, 2002). Of these activities, TFIIH kinase activity has been most closely linked to recruitment of the mRNA capping enzyme for transcript maturation (Rodriguez et al., 2000). In addition to pol II, the basal transcription factors TFIID, TFIIE, and TFIIF are phosphorylated by TFIIH in vitro (Ohkuma and Roeder, 1994).

Mounting evidence suggests that several proteins involved in activated transcription may be physiologically relevant substrates for TFIIH kinase activity. In particular, the transactivation capacity of several members of the nuclear receptor superfamily including retinoic acid receptor alpha [RARα (Rochette-Egly et al., 1997)], retinoic acid receptor gamma [RARγ (Bastien et al., 2000)], and estrogen receptor alpha [ERα (Chen et al., 2000)] have all been shown to be positively regulated by TFIIH phosphorylation in a response gene-dependent manner. Numerous other proteins involved in activated transcription have also been shown to be phosphorylated by TFIIH kinase in vitro. Of note, p53 interacts with and is phosphorylated by TFIIH in vitro (Ko et al., 1997; Lu et al., 1997), although it is not yet clear whether this modification modulates the transactivation function of p53, or its function in TFIIH-mediated nucleotide excision repair (Wang et al., 1996; Wang et al., 1995).

In addition to anchoring the kinase subcomplex to TFIIH (Busso et al., 2000; Sandrock and Egly, 2001), the precise functional role that Mat1 plays in the context of trimeric Cdk7-cyclin H-Mat1 or TFIIH remains poorly characterized. Mat1 appears to facilitate the assembly of the Cdk7-cyclin H-Mat1 trimeric complex (Devault et al., 1995; Fisher et al., 1995; Larochelle et al., 2001; Tassan et al., 1995) and to stabilize the complex in vivo (Rossi et al., 2001). Mat1 has also been implicated in determining the substrate specificity of Cdk7. For instance, p53 (Ko et al., 1997) and Oct transcription factors (Inamoto et al., 1997) are phosphorylated in vitro by Cdk7 in a Mat1-dependent manner. Moreover, the Mat1-dependent anchoring of the kinase subcomplex to TFIIH changes the substrate preference of Cdk7 from Cdk to CTD substrates (Rossignol et al., 1997; Yankulov and Bentley, 1997), and is a prerequisite for phosphorylation of at least TFIIE and TFIIF (Yankulov and Bentley, 1997).

We have previously reported that mice with a targeted inactivating allele of Mat1 undergo peri-implantation embryonic lethality which was coupled to an inability of mitotic lineages to proliferate (Rossi et al., 2001). Interestingly, we discovered that when taken to culture, Mat1-deficient embryos gave rise to viable post-mitotic trophoblast cells that could be maintained for a limited time in culture suggesting that transcriptional integrity of these cells was not severely compromised. In order to further investigate this possibility in vivo, we have conditionally disrupted Mat1 in both mitotic and post-mitotic adult lineages in the mouse using a tissue specific Cre/loxP approach.

The isolation and the restriction analysis of the Mat1 genomic locus have been previously reported (Rossi et al., 2001). A target vector was constructed in which a 3.8 Kb HindIII KpnI genomic fragment containing an exon corresponding to nucleotides 242-394 of the murine cDNA (DDJ/EMBL/GenBank accession No. U35249) was flanked by loxP sites. Cre-mediated deletion of this sequence yields an allele identical to the one previously described (Rossi et al., 2001), which was demonstrated to be a Mat1 null allele. Homologous recombination was screened by Southern blotting with 5′ and 3′ external probes where ten out of 2900 clones confirmed to be correctly targeted with both probes. PCR genotyping was achieved with the following primers: M10; 5′-GCC CTA TTT CAG GAG CCA GTC C, M12; 5′-TGA CCA AGC ATT TGT ATC TAT GAG CC, N4; 5′-GTC AGT TTC ATA GCC TGA AGA ACG. M10 and M12 amplify bands of 385 bp and 477 bp corresponding to the wildtype and flox alleles, respectively, while M10 and N4 amplify a 310 bp band corresponding to the null allele.

Crosses between Mat1^flox/+; KCN males and Mat1^flox/flox females were used to generate the experimental Mat1^flox/-; KCN animals, termed here mutant animals. As the Mat1^flox/+; KCN males were subjected to Cre recombination in the germline, they were found to only transmit the recombined null allele (-) to their progeny (n=739). The control littermates bearing varying genotypes (Mat1^flox/+, Mat1^flox/- and Mat1^flox/+ ;KCN) are phenocopies of each other and were thus collectively used as control animals.

For hematoxylin eosin (H&E) stained histology, tissues were fixed over night in 4% paraformaldehyde (PFA), dehydrated and embedded in paraffin. Sections (7 μm) were deparaffinized, rehydrated and stained. For plastic embedding, sciatic nerves were fixed in 1% PFA, 0.5% glutaraldehyde for 2 hours, treated with 1% osmium tetroxide (OsO₄), dehydrated in ethanol, and embedded in Epoxy resin. Semithin (1 μm) sections were stained with Toluidine blue (0.5% w/v) and boric acid (0.5% w/v). For transmission electron microscopy (TEM), ultrathin (60-90 nm) sections were cut on grids and stained with uranyl acetate and lead citrate. TEM samples were observed with a JEOL 1200EX electron microscope.

For immunofluorescence detection, sciatic nerves were embedded in O.C.T. and frozen on dry ice. Cryosections cut at 6 μm were fixed in 4% PFA (100% methanol for BrdU/Krox-20 double labeling) for 10 minutes, permeabilized with 0.2% Triton-X-100 in PBS, and blocked for 30 minutes with 5% normal goat serum (5% fetal bovine serum for BrdU/Krox-20 double labeling) in PBS with 0.2% Triton-X-100. Primary antibodies in blocking buffer were incubated for 3 hours, secondary antibodies for 45 minutes, washed with PBS and counterstained with Hoechst 33342 (Sigma; 0.5 μg/ml). Primary antibodies used were rabbit polyclonals against Krox-20 (1:50, CRP), Cre (1:1000, CRP), and Mat1 (1:100, FL-309, Santa Cruz), and sheep polyclonal against BrdU (1:100, Research Diagnostics). Secondary antibodies used were ALEXA 488 conjugated (1:500, Molecular Probes) or TRITC conjugated (1:500, Chemicon) anti rabbit, and ALEXA 488 conjugated anti sheep (1:500, Molecular Probes). Subsequent to Krox-20 staining, for BrdU detection, sections were denatured with 2 M HCl for 20 minutes prior to application of the anti BrdU primary antibody. All slides were mounted in Immumount (Shandon), viewed with Zeiss Axioplan 2 microscope, and documented with Zeiss Axiocam. Merge images were made with Zeiss Axio Vision version 3.0.6 SP3 software multichannel imaging.

β-galactosidase and alkaline phosphatase stainings on sciatic nerve cryosections were performed as previously described (Lobe et al., 1999) for sectioned tissues.

BrdU (B-5002; Sigma) was injected into animals intraperitoneally (50 μg/g body weight) once a day for 10 days in 0.9% NaCl, 7 mM NaOH solution. At day 11 sciatic nerves were collected and processed as described for immunofluorescence detection.

Mouse sciatic nerves were snap frozen in liquid nitrogen. Homogenization was performed in Laemmli Sample Buffer. Nerves were first cut into small pieces and then sonicated three times for 10 seconds with MSE Soniprep 150 at 18 micron amplitude on ice. Lysates were run on 10% SDS PAGE gels. Primary antibodies used were rabbit polyclonals against Mat1 (FL-309, 1:1000) and Cdk7 [1:000 (Mäkelä et al., 1995)], mouse monoclonal antibodies against CNPase (1:1000, Chemicon), and p44 (1:200, a kind gift from Pablo Valenzuela), and rat monoclonal against MBP (1:2500, Chemicon). Secondary antibodies used were horse radish peroxidase conjugated anti rabbit and anti mouse (both 1:5000, Chemicon) and anti rat (1:5000, Jackson Immuno Research Labs) antibodies.

A vector designed to conditionally disrupt Mat1 in a Cre/loxP-mediated manner was constructed and introduced into ES cells. In order to ensure that targeted region would yield an null allele (—) comparable to the allele generated in our previous study (Rossi et al., 2001), we targeted the identical 3.8 kb genomic region by flanking it with loxP sites (Fig. 1A). Ten ES cell clones (of 2900 screened) were confirmed to be correctly targeted by Southern blotting (Fig. 1B), and mice bearing the conditionally targeted allele (Mat1^flox) were generated.

Intercrossing Mat1^flox/+ and Mat1^+/- mice yielded F1 offspring bearing combinations of all the expected alleles in Mendelian ratios (N.K. and D.J.R., unpublished; Fig. 1C). The Mat1^flox/flox mice were healthy and fertile indicating that the conditional allele had no overt hypomorphic effects. To test the functionality of the flox allele in vivo, the Mat1^flox/flox conditional mice were crossed to a PGK-promoter-driven Cre-deleter strain (Lallemand et al., 1998) and the recombined Mat1 null allele (—) was observed in all offspring (Mat1^+/-).

To generate Mat1^-/- cell populations, the Mat1^flox/- animals were crossed to mice in which Cre recombinase was knocked-in to the Krox-20 locus [KCN; detailed characterization of the expression pattern in Voiculescu et al. (Voiculescu et al., 2000)].

As Krox-20 is known to be expressed in the male mitotic germ cells upon sexual maturation (Voiculescu et al., 2000), we were able to address the role of Mat1 in a mitotic cell population by following the fate of the germ lineage in targeted Mat1^flox/-;KCN males at various time points during sexual maturation. As sexual maturity in mice is typically reached at 6 to 7 weeks of age, mutant and control animals were analyzed at time points before sexual maturity (3 weeks postnatal), at sexual maturity (6 weeks postnatal) and after reaching sexual maturity (14 weeks postnatal).

As anticipated, at 3 weeks of age (P21; a time point prior to initiation of Cre expression in the testis) the developing germ lineage in the seminiferous tubules of mutant animals was completely indistinguishable from control animals (Fig. 2A; top panels). At 6 weeks postnatal however, pycnotic germ cells were readily observed in the seminiferous tubules in the testis of mutant animals (arrow in Fig. 2A). Moreover, an under-representation or absence of the cell types involved in spermatid differentiation was also observed (asterisk in Fig. 2A). These data were indicative of spermatogonia and spermatocyte cell death suggesting a requirement of Mat1 for germ lineage viability.

By 14 weeks postnatal (P100), the macroscopic appearance of the adult male reproductive organs from the mutant animals showed severe wasting and dramatically reduced cellularity compared to control animals (Fig. 2B). Examination of the seminiferous tubules in mutant animals revealed that they were entirely devoid of all germ cells and their derivatives (Fig. 2B, lower right panel). By contrast, both Sertoli cells in the seminiferous tubules and cells of the interstitium were present consistent with a lack of Krox-20 expression in these cell populations (Fig. 2B). In accordance with the histological analysis, mutant males generated in our experiments were found to be infertile.

Expression of the Krox-20 transcription factor in the peripheral nervous system is first noted at around embryonic day 15.5 and is required for the myelination of peripheral nerves (which commences at around the time of birth) where Krox-20 controls the expression of numerous genes involved in myelin synthesis (Topilko et al., 1994). Although myelinated Schwann cells are post-mitotic, they are nonetheless highly transcriptionally active both during the first month postnatal (when great quantities of transcripts encoding myelin structural components are generated), and after myelination is complete (Stahl et al., 1990).

The Krox-20-Cre animals used in our study have been used previously to target erbB2 for ablation wherein it was observed that greater than 98% of all peripheral myelinated Schwann cells were deleteriously affected shortly after birth resulting in a widespread peripheral neuropathy (Garratt et al., 2000). In marked contrast, our mutant animals were viable and healthy and displayed no signs of neuropathy or myelin dysfunction into early adulthood. To verify that the Krox-20-Cre allele used by Garratt et al., was working reproducibly in our animals we immunostained for expression of Cre recombinase and Krox-20 (as a marker for myelinated Schwann cells) in sciatic nerves of mutant and control 6-week-old animals (P42) (Fig. 3A). We observed robust expression of Krox-20 and Cre recombinase in comparable numbers of cells in the mutants while expression of the recombinase was absent in control (Mat1^flox/-) animals. These observations suggest that Cre recombinase was efficiently expressed, presumably in the myelinated Schwann cells. It should be noted that the sciatic nerve is a tissue comprised of several cell types and that myelinated Schwann cells represent an estimated 40-50% of the cells of the murine sciatic nerve (Garratt et al., 2000). We tested this assertion by counterstaining Krox-20 stained sciatic nerves with Hoechst and calculated that 46% (420/918) of all cells in control sciatic nerve and 49% (386/787) of the cells in the mutant sciatic nerve stained positive for Krox-20. Accordingly, 44% (159/359) of the cells in the mutant sciatic nerve stained positive for Cre. The other 50-60% of cells in the sciatic nerve include non-myelinated Schwann cells, endoneurial fibroblasts, and endothelial cells all of which do not express Krox-20 and therefore remain non-targeted for Mat1 ablation (the arrows in Fig. 3A indicate examples of such Krox-20/Cre-negative cells).

To confirm the efficiency of Cre-mediated recombination in the sciatic nerve, we crossed our animals to a transgenic strain bearing the Z/AP double reporter (Lobe et al., 1999). This enabled us to identify cells in which the Z/AP allele was either recombined (positive alkaline phosphatase (AP) staining) or non-recombined (positive β-galactosidase (lacZ) staining). We observed that a large fraction of the cells of the mutant sciatic nerve stained positive for AP indicating activity of the Cre recombinase (Fig. 3B). As expected, we also observed a significant fraction of cells which stained positive for lacZ highlighting the cells of the sciatic nerve which do not express Krox-20 or Cre and therefore remained non-recombined (Fig. 3B).

We then immunostained mutant and control sciatic nerves for expression of Mat1. While the sciatic nerve of the control animals stained positive for Mat1 in all cells, Mat1 immunoreactivity was only retained in the non-targeted cell types but was absent from the myelinated Schwann cells in the mutant sciatic nerve (Fig. 3C). Taken together, these results suggest that Mat1 was efficiently ablated in the myelinated Schwann cells of the sciatic nerve of Mat1^flox/-;KCN mutant animals.

Considering the proposed essential functions of Mat1, we were intrigued to find no over phenotype indicative of Schwann cell dysfunction in our young mutant mice following Mat1 disruption. We therefore examined the sciatic nerves from mutant and control animals at 1 month (P37) and 2 months (P67) of age (Fig. 4A). The sciatic nerves from mutant animals showed normal myelin thickness compared to axonal diameter. Moreover, the ratio between small and large diameter axons, and the density of myelinated axons appeared normal. Closer analysis of the morphology of the myelinating Schwann cells using transmission electron microscopy also failed to detect abnormalities in the 1 and 2 month old animals (N.K., unpublished). These results indicate that myelinating Schwann cells are capable of attaining a mature myelinated phenotype in the absence of Mat1.

Although the mutant animals did not show signs of Schwann cell dysfunction at an early age, at approximately 3 months of age they began to exhibit a variety of symptoms including gait abnormalities, and loss of body mass which was most pronounced in the hind limbs. These phenotypes were suggestive of neuropathy and subsequent neurogenic muscular atrophy. Histological analysis of the mutant sciatic nerves at this age (P97) indicated obvious signs of pathology in 64% (184/290) of the medium and large diameter axon/myelinated Schwann cell units (Fig. 4B). The most severe features included the presence of large diameter axons completely denuded of myelin (arrows in Fig. 4B), as well as large axioplasms surrounded by very thin myelin sheaths (arrowheads in Fig. 4B). By 5 months of age (P157) the phenotype of the mutant sciatic nerve was further exacerbated with essentially all (98%; 413/421) of the myelinated axons exhibiting morphological signs of pathology (Fig. 4B). Schwann cell death, myelin debris and denuded axons indicated extensive demyelination, and a plentitude of thinly sheathed axons suggested that remyelination was ongoing.

To study the pathological changes of this phenotype in greater detail transmission electron microscopy (TEM) was performed on the sciatic nerves from mutant and control animals. A representative field of the ultrastructure of a mutant nerve from a 5-month-old animal (P157) is shown in Fig. 5A, in which several medium diameter axons with thin myelin sheaths of varying thickness and reactive Schwann cell nuclei are noted. Macrophages (Fig. 5A) were also frequently observed presumably in response to myelin breakdown and Schwann cell death (Fig. 5B). Signs of ongoing remyelination were readily observed including denuded axons in association with pro-myelinating Schwann cells (Fig. 5B) and Schwann cells with very thin myelin sheaths (Fig. 5C). Non-myelinated Schwann cells with large reactive nuclei and enlarged nucleoli were also readily detectable suggesting possible transcriptional and proliferative activity (Fig. 5D). Higher magnification of remyelinating cells indicated that myelin synthesis and lamellae compaction was normal in the mutant as the major dense and intraperiod lines were comparable to control myelin (Fig. 5E,F). The myelin sheath in the remyelinating mutant nerve was not observed to reach full thickness as has been previously observed in other models of remyelination (Feldman et al., 1992; Gregson and Hall, 1973; Smith and Hall, 1988). Quantitation indicated a maximum of 15 lamellae per a medium diameter axon (3.0-3.5 μm) compared to an average of 49±2 in controls.

The extent of remyelination that we observed in the mutant sciatic nerve suggested that a proliferative response might underlie the remyelination phenotype. To test this possibility, in vivo BrdU labeling experiments were done on mutant and control animals of varying ages. Proliferation in the Schwann cell lineage was then assayed for by double immunofluorescence labeling of BrdU and Krox-20 on sciatic nerve sections (Fig. 6A). The merged image shown in Fig. 6A reveals BrdU labeled cells with green fluorescence, Krox-20 expressing cells with red fluorescence and BrdU/Krox-20 double-positive cells with yellow. Quantitation of proliferation of the Schwann lineage cells indicated that at approximately 1 month of age (P37) the percentage of BrdU/Krox-20 double-positive cells in the mutant (1.0%; 14/1412) was comparable to the control (1.0%; 14/1376) (Fig. 6B). At approximately 3 months of age (P97) however, the mutant exhibited a marked increase in Schwann lineage proliferation (6.2%; 79/1269). The near absence of proliferation in the Schwann lineage of the control at this age (0.1%; 1/1522) is normal and reflects the homeostasis of this post-mitotic lineage. Increased proliferation in the Schwann lineage was maintained in the mutant sciatic nerve at P157 with 4.2% (75/1795) of the cells scoring BrdU/Krox-20 double-positive (Fig. 6B). This was again in marked contrast to controls where no BrdU/Krox-20 double-positive cells (0/1407) were noted. The number of cells labeling exclusively for BrdU (and not for Krox-20) was similar at P37 between the control (2.6%; 36/1376) and the mutant (3.3%; 47/1412). After the onset of the demyelination phenotype however (at P97 and P157) the exclusively BrdU positive cell numbers were markedly elevated in the mutants (3.9%; 50/1269 and 9.7%; 174/1795, respectively) compared to the controls (1.2%; 18/1522 and 0.6%; 8/1407, respectively) (Fig. 6A, green cells in merge). This observation likely reflects proliferation of non-myelinated Schwann cells not yet expressing Krox-20 in addition to proliferation of macrophages, and endoneurial fibroblasts in response to demyelination.

Taken together these results indicate that after 3 months of phenotypic integrity, Mat1-deficient myelinating Schwann cells succumb to Mat1 loss resulting in a severe demyelination phenotype followed by proliferation in the Schwann lineage and remyelination.

In order to examine the biochemical properties of the mutant Schwann cells, sciatic nerves from mutant and control animals were dissected before the onset of the demyelinating phenotype (at P70) and subjected to western blotting analysis (Fig. 7). In accordance with our immunofluorescence and double reporter data, we found that Mat1 levels were greatly reduced in the mutant lysates compared to controls with remaining Mat1 immunoreactivity attributable to the nontargeted cells. Cdk7 expression levels were also found to be reduced in the mutant, while the levels of the core TFIIH subunit p44 remained unchanged in the mutant (Fig. 7). These results presumably reflect a destabilization of the trimeric CAK complex in the absence of Mat1 as has been previously reported (Rossi et al., 2001), while the core TFIIH complex remains stable.

We then assayed for expression of the myelinated Schwann cell markers CNPase (Sprinkle, 1989) and myelin basic protein (MBP) and found that expression of both these proteins was fully maintained in the mutant compared to the control sciatic nerve (Fig. 7). Of note, the expression levels and ratio of the three major MBP isoforms found in adults were comparable in mutants and controls. Coupled with our histological data, these results indicate that Mat1-deficient Schwann cells are able to obtain and then maintain a mature myelinated phenotype into adulthood after the genetic ablation of Mat1.

Mat1 is believed to be required for cell cycle progression (as a component of the Cdk-activating kinase), and RNA polymerase II-mediated transcription (as a component of TFIIH). As both of these functions are essential for the viability of cells, it has therefore proven difficult to dissect the role of Mat1 in these two functional pathways in vivo. The unexpected differential requirement for Mat1 in embryonic mitotic and post-mitotic cells in culture (Rossi et al., 2001) prompted us to generate Mat1 loxP conditional mice where Mat1 was targeted for ablation in both mitotic and post-mitotic cells in vivo following crossing with a Krox-20-Cre knock-in strain.

Ablation of Mat1 in mitotic cells was analyzed in the testis, where Krox-20 is expressed in the mitotic germ cells upon sexual maturation. It was noted that this cell population was rapidly rendered non-viable upon ablation of Mat1 indicating that the requirement of Mat1 for mitotic lineage survival is not restricted to embryonic lineages (Rossi et al., 2001). These phenotypes in both embryonic and adult mitotic lineages are consistent with defects in cell cycle progression, which may reflect a deficit in Cdk activation. It is also interesting to note that although the exact pattern of Krox-20 expression in the development of male germ cells has not been characterized, the absence of spermatogonia stem cells (in addition to all other germ cells) in the mutant adult seminiferous tubules suggests that recombination of Mat1 happens at a very early stage of spermatogenesis during the generation of type A spermatogonia stem cells.

The Krox-20-Cre strain also allowed us to disrupt Mat1 before the onset of myelination in the post-mitotic myelinating Schwann cells of the peripheral nervous system. We reasoned that this approach might allow us to circumvent any potentially essential functions of Mat1 in Cdk activation/cell cycle progression and thus allow us to address the role of Mat1 in a strictly transcriptional context. During the first postnatal month the myelinating Schwann cells of the peripheral nerves of rodents express large amounts of proteins involved in myelin synthesis and compaction including P0, MAG, PLP, CNPase, and MBP (Stahl et al., 1990). After myelination is complete, ongoing transcriptional activity is required for Schwann cell viability in order to maintain metabolism, as well as to preserve myelin structure and integrity. Indeed, several studies have shown that inhibition of RNA polymerase II-mediated transcription in Schwann cells in vivo leads to rapid cellular degeneration and death in a few days (Benavides and Alvarez, 1998; Court and Alvarez, 2000; England et al., 1988). We found that while post-mitotic Schwann cells required Mat1 for long-term survival, they were nonetheless fully capable of completing the transcriptional program essential for the myelination of peripheral axons. Moreover, we found that these cells remained morphologically normal and the mice showed no physiological evidence for Schwann cell dysfunction up to 3 months of age. These results argue that Mat1 is not widely required for RNA polymerase II-mediated transcription in post-mitotic mammalian cells. This is in accordance with our previous work demonstrating that cultured post-mitotic embryonic trophoblast cells remained transcriptionally active in the absence of Mat1 (Rossi et al., 2001).

Our results significantly differ from several genetic studies in Saccharomyces cerevisiae utilizing temperature sensitive mutants of any of the three TFIIH kinase subunits Kin28, Ccl1 or Tfb3 (Faye et al., 1997; Holstege et al., 1998; Valay et al., 1996; Valay et al., 1995). In these studies it has been found that the vast majority of pol II transcription was terminated (within 10-20 minutes) upon switching to the restrictive temperature and the viability of the yeast was rapidly lost. In contrast, other lines of evidence suggest a less predominant role for TFIIH kinase in budding yeast. For example, mutations eliminating either the CTD or TFIIH kinase do not appear to inhibit the in vivo transcription of all genes (Holstege et al., 1998; Lee and Lis, 1998; McNeil et al., 1998). Moreover, a study utilizing a strain of yeast in which a predicted kinase-deficient KIN28 mutant was introduced into the endogenous KIN28 locus was shown to be viable (albeit slow growing) despite showing no detectable TFIIH kinase activity or CTD phosphorylation (Keogh et al., 2002).

Several lines of evidence support the suggestion that mammalian TFIIH kinase activity may not be as widely required for pol II-mediated transcription as has been suggested in yeast systems. In vitro assays with purified human factors have shown that a CTD-free pol II (Buratowski and Sharp, 1990; Zehring and Greenleaf, 1990), a kinase-deficient TFIIH (Mäkelä et al., 1995) or TFIIH without the kinase subunit (Rossignol et al., 1997; Tirode et al., 1999) are functional in both basal and activated transcription. In studies where the mammalian SMCC coactivator was found to operate independently of TFIIH in transcription activation it was suggested that one or more of the other CTD kinases might subserve normal TFIIH kinase functions (Gu et al., 1999). Moreover, it has also been shown that transcription of heat shock genes does not appear to require TFIIH kinase activity (Dubois et al., 1997; Egyhazi et al., 1998). The dispensability of TFIIH kinase in these systems however is in marked contrast to the core TFIIH subunits which are absolutely essential for RNA pol II-mediated transcription (Tirode et al., 1999). In support of this, mice with a homozygous ablation of the XPD helicase subunit of TFIIH die at the 2-cell stage of embryonic development (de Boer et al., 1998), while mice homozygous for a disruption of Mat1 survive up to implantation and give rise to viable post-mitotic cells which can be further maintained in culture (Rossi et al., 2001).

Our results indicate that myelinated Schwann cells are ultimately dependent on Mat1 for survival. Although we cannot exclude that this reliance is due to a Cdk-activation deficit, a defect in nucleotide excision repair, or a yet uncharacterized function of Mat1, we think it more likely that this phenotype is due to a deficit in the production of a subset of transcripts. Such a subset may include target genes transactivated by the nuclear hormone receptors known to be regulated by TFIIH kinase activity (Bastien et al., 2000; Chen et al., 2000; Keriel et al., 2002; Rochette-Egly et al., 1997). Interestingly, numerous hormones such as estradiol, testosterone and progesterone have been shown to be active through their cognate receptors in Schwann cells (Mercier et al., 2001; Schumacher et al., 2001). It will be interesting to determine whether the activity of any of these proteins might be modulated by TFIIH kinase and whether the lack of such modifications would lead to changes in target gene responsiveness.

The myelinated Schwann lineage of the peripheral nervous system has been shown to have a great capacity for regeneration in response to injury. The two types of Schwann cells implicated in this regenerative capacity are the myelinated Schwann cells and the non-myelinated Schwann cells. In the case of the myelinated Schwann cells, certain injuries trigger a dedifferentiation, proliferation and subsequent redifferentiation cycle known as Wallerian degeneration (reviewed in Stoll and Muller, 1999). The role of non-myelinated Schwann cells in injury models however is less clear although proliferation of non-myelinated Schwann cells has been observed which might then presumptively differentiate and take on a myelinating Schwann cell fate (Griffin et al., 1990; Messing et al., 1992). As the insult leading to demyelination in our model is the result of a genetically programmed death of the myelinated Schwann lineage, we therefore suggest that the non-myelinated Schwann cell pool likely underlies the remyelination that we see in our mutant animals. These animals therefore provide an ideal system in which to study the properties of non-myelinated Schwann cells in remyelination.

We thank Peter Lonai for the PGK-Cre mice, and Corrinne Lobe and Andras Nagy for the Z/AP mice. Jorma Toppari is acknowledged for consulation. Birgitta Tjäder and Susanna Räsänen are acknowledged for excellent technical assistance. This study was supported by grants from The Finnish Medical Foundation, Finnish Cancer Organization, Sigrid Juselius Foundation, and Academy of Finland. N.K. is a graduate student at the Helsinki Graduate School of Biotechnology and Molecular Biology, and D.J.R. is a graduate student at the Helsinki Biomedical Graduate School.