Abstract
In peripheral nerves, large caliber axons are ensheathed by myelin-elaborating Schwann cells. Multiple lines of evidence demonstrate that expression of the genes encoding myelin structural proteins occurs in Schwann cells in response to axonal instructions. To gain further insight into the mechanisms controlling myelin gene expression, we used reporter constructs in transgenic mice to search for the DNA elements that regulate the myelin basic protein (MBP) gene. Through this in vivo investigation, we provide evidence for the participation of multiple, widely distributed, positive and negative elements in the overall control of MBPexpression. Notably, all constructs bearing a 0.6 kb far-upstream sequence, designated Schwann cell enhancer 1 (SCE1), expressed at high levels in myelin-forming Schwann cells. In addition, robust targeting activity conferred by SCE1 was shown to be independent of otherMBP 5′ flanking sequence. These observations suggest that SCE1 will make available a powerful tool to drive transgene expression in myelinating Schwann cells and that a focused analysis of the SCE1 sequence will lead to the identification of transcription factor binding sites that positively regulate MBPexpression.
- myelin basic protein
- Schwann cells
- myelination
- MBP regulation
- MBP regulated transgenes
- sensory and motor fibers
- Krox-20
Schwann cells arise from a common pool of neural crest cells and differentiate into myelinating or nonmyelinating glia (Mirsky and Jessen, 1996). Axonal signals control the decision to myelinate, quantitative features of the sheath, and are required for its subsequent maintenance (Aguayo et al., 1976a,b;Berthold, 1978). These axonal instructions operate, in part, by controlling the expression of the genes encoding myelin structural proteins. In PNS myelin these include protein zero (P0), peripheral myelin protein 22 (PMP22), and myelin basic protein (MBP). High-level expression of these genes coincides with myelin elaboration, continues throughout myelinogenesis, and ceases should the axon be disrupted (Gupta et al., 1988; Trapp et al., 1988; Lamperth et al., 1990; LeBlanc and Poduslo, 1990; Stahl et al., 1990; Snipes et al., 1992). Molecules implicated in the control of Schwann cell maturation and myelin formation include hormones such as progesterone, the second messanger cAMP, the zinc finger transcription factor Krox-20, and the POU domain transcription factor Tst-1/SCIP/Oct-6 (SCIP) (Lemke and Chao, 1988; Monuki et al., 1989, 1990; He et al., 1991; Morgan et al., 1991;Topilko et al., 1994; Koenig et al., 1995; Weinstein et al., 1995;Bermingham et al., 1996; Jaegle et al., 1996; Zorick and Lemke, 1996). However, neither the molecules that transduce axonal instructions nor the Schwann cell signaling pathways that they control are known. Similarly, the most downstream components of such pathways, the transcription factors, and DNA regulatory elements that confer expression to myelin genes remain to be elucidated.
Previously, the expression of reporter genes promoted by one to several kilobases of 5′ flanking sequences from the P0, MBP, and CNPase genes were investigated in transgenic mice. The constructs incorporating P0 orCNPase sequence expressed in Schwann cells indicating the presence of positive Schwann cell targeting elements (Messing et al., 1992; Gravel et al., 1998). In contrast, constructs incorporating similar lengths of MBP 5′ flanking sequence were expressed only in oligodendrocytes, the myelin-forming cell type in the CNS, leaving the location ofMBP-related Schwann cell targeting elements unknown (Foran and Peterson, 1992; Gow et al., 1992; Miskimins et al., 1992;Goujet-Zalc et al., 1993; Stankoff et al., 1996). Here, using the same approach, we searched within previously uncharacterized regions of theMBP locus for Schwann cell regulatory functions. Sequences contributing both positively and negatively to theMBP expression phenotype were found widely distributed across 12 kb of MBP 5′ flanking sequence. We show that one of these, a 0.6 kb Schwann cell enhancer 1 (SCE1), confers high-level reporter gene expression to myelin-forming Schwann cells. Furthermore, we demonstrate that SCE1 is a classic enhancer of transcription, capable of conferring expression through a heterologous promoter and in an orientation- and position-independent manner. Because SCE1 functions robustly in isolation from further MBP promoter sequence, the search forcis-regulatory elements that positively regulateMBP expression in Schwann cells can be significantly focused within this 0.6 kb.
MATERIALS AND METHODS
Generation of reporter constructs. A 15 kbBamHI MBP genomic fragment, containing ∼12.0 kb of MBP 5′ flanking sequence, was obtained by screening a lambda DASH-129 mouse genomic library (J. Rossant, Mount Sinai, Toronto, Ontario, Canada). This sequence was subcloned in theBamHI site of pSK (B clone). A SacII (−9.0 kb)/XbaI (−3.1 kb) fragment from B clone was inserted into the respective sites in clone pm12 (3.1 kb MBP 5′ flanking sequence in XbaI/XmaI sites of pSK−;Foran and Peterson, 1992) to generate the −9.0 kb MBPpromoter (clone 8). B clone was digested with SacII followed by intramolecular ligation of the SacII ends to generate the −12.0 kb (BamHI) to −9.0 kb (SacII)MBP 5′ flanking sequence in pSK (B subclone 1A) (all vectors other than pm12 from Stratagene, La Jolla, CA; restriction enzymes from New England BioLabs, Mississauga, Ontario, Canada).
To generate reporter constructs, d10 lacZ (Foran and Peterson, 1992) was released from a pUC18 subclone with SalI and cloned in the SalI site of clone 8 (3′ to the −9 kbMBP promoter). This clone contained a double insert oflacZ, and the second insert was released byBamHI digestion followed by intramolecular ligation (clone 5). Constructs containing 9.0 kb (SacII), 8.5 kb (NaeI), 7.0 kb (SphI), or 6.0 kb (KpnI) of MBP promoter were obtained by restriction digestion of clone 5 and agarose (Boehringer Mannheim, Laval, Quebec, Canada) gel purification (0.5% Tris-Acetate-EDTA). A KpnI/BamHI fragment of clone 5 (containing −6.0 kb MBP-lacZ) was cloned into the respective sites in B subclone 1A to generate a clone containing the −12.0 to −9.0 5′ MBP fragment at the 3′ end of lacZ in 5′ to 3′ orientation. This clone was digested with KpnI/SacII and the construct was gel purified as described above. Two constructs were used to test the position and orientation independence of SCE1. SCE1 (0.6 kb;SacII/SacI) was isolated from clone 5 bySacI digestion and cloned in the SacI site of pSK+ (clone 6). We isolated −6.0 kbMBP-lacZ from clone 5 (KpnI/BamHI) and cloned it into the same sites in clone 6 to generate −6.0MBP-lacZ-5′(SCE1)3′. The construct was obtained by linearizing with KpnI. To test the reverse orientation, SCE1 was isolated from clone 5 (SacI) and cloned in the SacI site of pBS (clone 7). We isolated −6.0MBP-lacZ from clone 5 withKpnI and cloned it into the KpnI site of clone 7 to generate −6.0 MBP-lacZ-3′(SCE1)5′. The construct was released with SphI/SacII for pronuclear injection. A clone containing SCE1 and −3.1 kbMBP-lacZ also was generated. We isolated 3.1 kb MBP-lacZ from clone pm12 withXbaI and cloned it into the XbaI site of clone 7, resulting in a clone containing 5′(SCE1)3′-−3.1 kbMBP-lacZ). The construct was released with SacII/SphI.
To generate SCE1-hsp-lacZ constructs, the minimal 0.3 kb hsp68 promoter (HindIII/NcoI) ligated tolacZ (clone p610ZA; R. Kothary, University of Ottawa, Ottawa, Ontario, Canada) was used. The 0.6 kb SCE1 (SacII/SacI) was blunted (Klenow; Boehringer Mannheim) and inserted in the EcoRV site of pKS+. In clone KS-SCE 8A, the 5′ end of the enhancer is closest to theSacII site of the pKS+ multiple cloning site (and the 3′ end closest to the KpnI site). We isolated 0.3 kb hsp68-lacZ from clone p610ZA (HindIII/KpnI) and cloned it into the same sites in KS-SCE 8A to generate SCE-hsp 2G (5′(SCE1)3′-hsp68-lacZ in pKS+). The construct was released by SmaI digestion and purified as above. To generate a construct having the SCE1 in 3′ to 5′ orientation, the 0.6 kb SCE1 (SacII/SacI) was cloned in theSacII/SacI sites of pSK+ (clone SK-SCE). We isolated 0.3 kb hsp68-lacZ from p610ZA (HindIII/KpnI) and cloned it into the same sites in SK-SCE (clone SCE-hsp 1B: 3′(SCE1)5′-hsp68-lacZ). This construct was linearized with KpnI.
To test whether the Krox-20 site within SCE1 is essential for enhancer function, a 5 bp mutation that abolishes Krox-20 binding [(−)GCGTGGGTG→GCGGTTTCG; Chavrier et al., 1990; Forghani et al., 1999] was introduced into SCE1, and the mutated SCE1 (δKrox-20-SCE1) was ligated to the 3′ end of the lacZ reporter gene driven by the 6 kbMBP promoter. To generate δKrox-20-SCE1, the 0.6 kb SCE1 (SacII/SacI) was blunted with Klenow (Boehringer Mannheim) and inserted in the blunted (Klenow)XbaI site of pKS+. In clone KS-SCE 11A, the 5′ end of the enhancer is closest to the KpnI site of the pKS+ multiple cloning site (and the 3′ end closest to the SacII site). To introduce a Krox-20 mutation into SCE1, complementary oligonucleotides encoding SCE1 sequence flanked by MluI and Sbf I sites and bearing the mutated Krox-20 site (custom made by Sheldon Biotechnology Centre, McGill University, Montreal, Quebec, Canada) were annealed and cloned into their respective sites within clone KS-SCE 11A, replacing the wild-type sequence and resulting in a mutant 0.6 kb SCE1 (δKrox-20-SCE1; confirmed by sequencing of both strands). Then, the final clone was generated by isolating the −6.0 kbMBP-lacZ from clone 5 (KpnI/BamHI) and cloning it into theKpnI/BamHI sites of pKS-δKrox-20-SCE1. The construct was released by KpnI/SacII digestion.
Derivation of transgenic mice. Transgenic mice were derived by injection of DNA into the pronuclei of B6C3F2 zygotes as previously described (Foran and Peterson, 1992). Injected zygotes were transplanted into the oviducts of B6C3F1 females rendered pseudopregnant by mating with vasectomized males. Litters were delivered either spontaneously or by cesarean section 18 d later. Primary transgenic mice and mice derived from established lines were investigated for transgene expression (Table1).
Transgenic mice used for the functional analysis of theMBP promoter
Histochemical detection of β-galactosidase activity. Mice were anesthetized with avertin intraperitoneally and perfused transcardially with 0.5% paraformaldehyde and 2.5% glutaraldehyde in 0.1 m phosphate buffer, pH 7.4, all at 4°C. Tissues to be analyzed were recovered and post-fixed for an additional hour. Samples were then rinsed in 0.1m phosphate buffer and incubated as whole mounts at 37°C for various times ranging from <1 hr to overnight in stain consisting of 2 mm MgCl2, 5 mmK3Fe(CN)6, 5 mmK4Fe(CN)6 and 0.4 mg/ml Bluo-gal (Life Technologies, Burlington, Ontario, Canada). In some preparations, detergents sodium deoxycholate and NP-40 were added to the stain at 0.01 and 0.03%, respectively, to permeabilize the tissue, thus assisting the penetration of β-galactosidase substrate. After the whole-mount histochemical reaction, some samples were processed for plastic embedding. Typically, tissue was osmicated before dehydration and embedding in Epon. After polymerization, the blocks were sectioned at 1 μm, and the sections were mounted on slides and viewed either directly or after staining with toluidine blue. Additional tissue was cryoprotected by immersion in 30% sucrose before freezing, and 12 μm cryostat sections were made and subsequently incubated in stain containing 0.8 mg/ml X-gal (Life Technologies). To investigate prenatal expression, we applied a histochemical technique capable of detecting low-level β-galactosidase activity in sections. Fetuses were recovered and immersion-fixed for 1 hr in the same aldehyde mix at 4°C. To cryoprotect, they were then incubated at 4°C in 30% sucrose overnight before freezing in isopentane precooled in liquid nitrogen. Cryostat sections, 12 μm thick, were picked up on slides and dried for 30 min at room temperature and 20 min at 37° C. Sections were then post-fixed by immersion in 4% formalin and 7.5% sucrose in 0.1 m phosphate buffer for 20 min at 4°C, washed for 5 min in buffer alone three times, then transferred to β-galactosidase stain (X-gal at 0.8 mg/ml) and incubated at 37° C overnight. Slides were then coverslipped using 10% glycerol as the mounting medium.
RESULTS
In a search for the cis-regulatory sequences that control MBP expression in Schwann cells, we analyzed the expression conferred to MBP-regulated reporter constructs in transgenic mice. In the initial constructs we evaluated,lacZ was promoted by ∼6, 7, 8.5, or 9 kb ofMBP 5′ flanking sequence (Fig.1). Constructs containing flanking sequences extending to −8.5 kb expressed only in oligodendrocytes, whereas the construct regulated by the 9.0 kb sequence also expressed in Schwann cells (Figs. 1, 2), indicating that one or more Schwann cell targeting elements are located in the sequence between −9 and −8.5 kb.
MBP-promoted lacZ reporter constructs used to map Schwann cell enhancers and repressors in transgenic mice. We cloned 12 kb of MBP 5′ flanking sequence from a lambda DASH-129 mouse genomic library. Constructs incorporating various lengths of these MBP sequences were then used to derive transgenic mice, and transgene expression was assessed by the β-galactosidase histochemical assay. 5′ flanking sequences extending to −6, −7, −8.5, and −9 kb were ligated to the lacZ reporter gene, and the sequence lying between −9 and −8.5 kb was found to be required for Schwann cell expression. To evaluate potential effects of position and orientation, the slightly longer 0.6 kb fragment extending from −9.0 to −8.4 kb, designated SCE1, was ligated to constructs promoted by −3.1 or −6 kb MBP sequence. A second Schwann cell targeting activity was detected in the further upstream sequence (−12 to −9 kb), and because it also functions as an enhancer when ligated 3′ of the lacZ reporter, it was designated SCE2. The striped region in the −6 kbMBP-lacZ-5′(SCE1)3′ construct designates the pSK+ vector included in the injected construct.
β-galactosidase labeling in transgenic mice bearing MBP-promoted, SCE-containing constructs. In transgenic mice bearing MBP 5′ flanking sequences that include SCE1, both oligodendrocytes and Schwann cells label intensely with incubation times varying from minutes to hours. a,Lumbar spinal cord and attached roots from a 21-d-old −9 kb line 17 mouse. Inset, A 1-μm-thick cross section of L4 ventral root embedded in plastic subsequent to whole-mount labeling. Bluo-gal reaction product is associated with the majority of the myelin and Schwann cell cytoplasmic profiles. b, A 21-d-old mouse bearing a −6.0 kb MBP-lacZ construct with SCE2 ligated to the 3′ end of lacZ expresses β-galactosidase in both oligodendrocytes and Schwann cells.Inset, A 1-μm-thick cross section of whole-mount-stained L4 root demonstrates labeling of a subpopulation of myelin and Schwann cell profiles in this mosaic, primary transgenic mouse. c, Reporter constructs bearing the −6 kb MBP promoter target expression to oligodendrocytes but not Schwann cells. d, Teased fiber preparation from an SCE1-bearing transgenic mouse shows β-galactosidase reaction product accumulated in the cytoplasmic compartments of the Schwann cell, including the perinuclear cytoplasm (asterisk), Schmidt–Lantermann incisures (closed arrowhead), and paranodal loops (open arrowhead). e, A 12-μm-thick cryostat cross section from a sciatic nerve of a transgenic mouse bearing the SCE1-containing −9 kb promoter shows labeling on many myelin and Schwann cell profiles. Because the intra-Schwann cell distribution of the reaction product is not uniform along the internode (d), not all such profiles should be labeled. Scale bar: a–c, 3.1 mm; a, b, insets, 0.05 mm; c, 0.05 mm; d,0.17 mm; e, 0.05 mm.
To assess whether this distal sequence had functions characteristic of an enhancer, we generated constructs in which the slightly longer −9 to −8.4 kb sequence was ligated, in both orientations, to the 3′ end of the lacZ reporter promoted by 6 kb ofMBP 5′ flanking sequence (Fig. 1). In another construct, it was ligated immediately upstream of 3.1 kb of MBP 5′ flanking sequence driving lacZ. Consistent with classic enhancer function, robust expression was observed in the PNS of multiple lines bearing all three constructs (Table 1). The Schwann cell targeting conferred by this 0.6 kb MBP sequence was assigned the interim designation SCE1.
To search for further sequences capable of conferring Schwann cell expression, the 3 kb sequence immediately upstream of SCE1 (−12.0 to −9.0) was ligated to the 3′ end of the lacZ reporter gene, driven by 6 kb of MBP 5′ flanking sequence (Fig. 1). So far, one transgenic mouse that failed to transmit the transgene has been analyzed. Although expression of the transgene was mosaic, those Schwann cells that labeled did so intensely (Fig. 2, Table 1). Because the MBP sequence in this construct was ligated 3′ of lacZ, its provisional targeting ability also appeared to be mediated through enhancer activity and it was assigned the interim designation SCE2. The sequence extending from −12 to −8.4 kb (including both SCE1 and SCE2) has been deposited to GenBank (accession number AF277397).
The observation that SCE1 functions in a position- and orientation-independent manner was made in the context of 6 or 3.1 kb of proximal MBP 5′ flanking sequence. Consequently, Schwann cell targeting activity could require interaction between SCE1 and additional elements contained within the proximalMBP promoter. To investigate this possibility, the −9.0 to −8.4 kb sequence was ligated, in both orientations, 5′ of a heterologous promoter (0.3 kb hsp68) driving the lacZ reporter gene. Transgenic mice bearing this SCE1-hsp68-lacZ construct were derived, and 7 of 10 independent transgenic lines, including those bearing SCE1 in either orientation, expressed β-galactosidase in myelin-forming Schwann cells (Fig. 3, Table 1). Because these constructs were expressed at high levels in the myelinating Schwann cells of multiple lines, SCE1 contains robust Schwann cell enhancer activity that is sufficient to target Schwann cell expression.
SCE1 targets Schwann cell expression in the context of a heterologous promoter. a, SCE1 was ligated to a 0.3 kb hsp68 minimal promoter in either orientation, and each construct was used to derive transgenic mice. The sequence represented by the striped box in 3′(SCE1)5′-hsp68-lacZ is from the vector, pSK+.b, Dorsal view of a whole-mount histochemical β-galactosidase preparation from a 7-d-old SCE1-hsp68-lacZ (line 18) mouse showing the intense labeling of fibers in lumbar spinal roots but no expression in oligodendrocytes in the CNS. c, d, Cranial nerves and cervical spinal roots also demonstrate intense labeling, whereas neither oligodendrocytes nor other cell types label in the brain. Note that this specimen was a B6C3F1 derivative and was pigmented accounting for the black pigmentation, and not reaction product, observed in the eyes. Scale bar: b, 1.6 mm; c,d, 4.4 mm.
To determine how closely the expression conferred by SCE1 tracks the expression phenotype of the endogenous MBP gene, we analyzed prenatal and postnatal transgenic mice bearing SCE1-containing constructs. The endogenous MBP locus is expressed at low but detectable levels in the developing PNS of prenatal mice (Bachnou et al., 1997). Transgenic mice bearing SCE1-hsp68-lacZ constructs similarly express low but detectable levels of β-galactosidase activity in fetal peripheral nerves from embryonic day 14 (E14) through birth (Fig.4a,b). Also reflecting the endogenous expression program, high-level expression of β-galactosidase appeared in Schwann cells coincident with myelin formation when accumulation of endogenous MBP mRNA markedly increases. As predicted by the temporally discordant myelination programs in dorsal and ventral spinal roots of mice (Baron et al., 1994), ventral roots began to label intensely 1–2 d before dorsal roots (Fig.4c,d). During the first postnatal week of mouse development, the number of myelinating Schwann cells greatly increases (Bray et al., 1977), and the apparent level of β-galactosidase accumulation, assessed by the rapidity and intensity of histochemical labeling, increased in parallel. Because similar expression programs were observed for SCE1 containing constructs promoted either by hsp68 or proximal MBP sequence, we conclude that this short enhancer sequence is capable of directing the major features of the endogenous MBP expression program during Schwann cell maturation.
Developmental expression of transgenes bearing SCE1. Low-level transgene expression is detected in some regions of the PNS in an SCE1-hsp68-lacZ line 18 fetus at E15.5. Mixed nerves exiting the spinal column have detectable but unevenly deposited β-galactosidase reaction product (a), whereas the trigeminal nerve is more obviously and uniformly labeled (b). High-level expression, detectable in whole-mount preparations, initiates only in the postnatal period. c, Whole-mount preparation of a 2-d-old −9 kb MBP-lacZ mouse (line 17) shows labeling in ventral but not dorsal spinal roots, reflecting the relative developmental delay in the myelination program of dorsal roots. d, Electron micrograph prepared from whole-mount-labeled tissue reveals β-galactosidase reaction product deposited adjacent to developing myelin sheaths in the L4 ventral root. At this stage of development, whereas some ventral root fibers have initiated sheath formation, the unlabeled dorsal roots (data not shown) do not contain any myelin profiles. e, Whole-mount preparation of roots and dorsal root ganglia from a preweaning −9 kbMBP-lacZ transgenic mouse. Uniform and intense staining is observed in both dorsal and ventral roots, whereas neither the neurons nor non-Schwann cell glia in the dorsal root ganglia label. f, In 3-month-old −9 kb line 17MBP-lacZ transgenic mice, Schwann cells ensheathing motor axons in ventral roots no longer label, whereaslacZ expression continues, at readily detectable levels, in Schwann cells ensheathing sensory axons in dorsal roots (root modality was identified by the spinal cord insertion site). Scale bar:a, b, 0.18 mm; b, 0.18 mm;c, 1.14 mm; d, 4 μm; e,1.25 mm; f, 1.6 mm.
Despite the close relationship between the expression in peripheral nerves of the transgene and the endogenous MBP locus, an apparent discrepancy was observed during midfetal development. Of the three SCE1-hsp68 lines examined (17, 18, and 42), all three consistently revealed transient β-galactosidase expression in a small subpopulation of cells in the neural tube, brain, and retina (data not shown). Because constructs bearing SCE1 with the 6 kbMBP promoter are not similarly expressed (our unpublished observations), putative negative regulatory elements serving to repress MBP expression in these cell populations appear to be located within the first 6 kb ofMBP 5′ flanking sequence. Further ectopic expression attributable to elements within SCE1 was not observed.
Throughout maturity, a stable level of MBP mRNA accumulates in peripheral nerves (LeBlanc and Poduslo, 1990; Stahl et al., 1990;Snipes et al., 1992). Reflecting this endogenous program, SCE1-hsp68-lacZ lines (17, 49, and 54) maintain expression throughout maturity (examined up to 7–9 months of age). In contrast, whereas intense labeling was observed in lines 18 and 42 during preweaning development, by 3 months of age only faint staining could be elicited. Such differential regulation could arise from unusual effects of transgene integration sites. Alternatively, comparison of transgene expression in mice from ongoing C57BL/6 and C3H backcross programs consistently revealed greater PNS labeling intensity in those backcrossed to C57BL/6 (data not shown). Consequently, the unique genetic background of each B6C3F1 derived transgenic line might be a contributing factor in maintaining or downregulating the mature expression phenotype.
As mice bearing the 9.0 kb promoted construct matured, there was a modest decline in β-galactosidase labeling intensity throughout their mixed nerves (lines 17 and 32) and in the mosaic line 24, expression ceased. Surprisingly and consistently, in mice >3 months of age from lines 17 and 32, dorsal roots continued to label, but a further precipitous decline in transgene expression was observed in ventral roots (Fig. 4f). In line 17 mice, expression appeared to be shut off, whereas in line 32, ventral root downregulation was not absolute, but labeling intensity was clearly weaker than that observed in dorsal roots. A similarly dramatic difference in dorsal and ventral root labeling intensity was observed in one line of SCE1-hsp68-lacZ mice (also designated 17). These combined observations suggest a novel level of heterogeneity among Schwann cells that coincides with the modality (sensory versus motor) of the innervating axon. Although a striking finding in the multiple lines where it was observed, this phenomenon was not encountered in all SCE1-bearing lines. Notably, both dorsal and ventral roots in mice from the SCE1-hsp68-lac Z lines 49 and 54 continued to label at an apparently similar level throughout maturity.
Because our studies demonstrated that the 0.6 kb SCE1 sequence contained the elements necessary for targeting high-level Schwann cell expression during maturation, we evaluated the sequence for known regulatory elements that could control MBP expression. We searched for such elements using MacVector software, as well as the online TFSEARCH program supported by TRANSFAC databases (Heinemeyer et al., 1998). Among the >150 putative elements recognized in the 0.6 kb sequence, one with a perfect match to a Krox-20 binding site was encountered near the 3′ end of SCE1 [(−)GCGTGGGTG; Sham et al., 1993; Fig. 5a]. In mice homozygous for null Krox-20 alleles, Schwann cells ensheath axons but fail to elaborate myelin (Topilko et al., 1994), demonstrating that Krox-20 plays an essential role in Schwann cell maturation or myelination. Because the Krox-20 element within SCE1 is located within the 0.1 kb region of overlap with the 8.5 kb promoter and transgenic animals bearing the 8.5 kb promoted construct do not express in Schwann cells (Table 1), we conclude that the Krox-20 element is not sufficient for conferring Schwann cell expression. To determine whether it is necessary for SCE1 enhancer activity, perhaps through interactions with additional SCE1 elements, we introduced a 5 bp mutation [(−)GCGTGGGTG→GCGGTTTCG that abolishes Krox-20 binding; Chavrier et al., 1990; Forghani et al., 1999]. A construct-bearing SCE1 with this mutant Krox-20 site (δKrox-20-SCE1) at the 3′ end of lacZ and driven by the 6 kbMBP promoter was generated (Fig. 5, Table 1). To date, transgenic mice from one line bearing the δKrox-20-SCE1 construct have been analyzed [at postnatal day 30 (P30) and P160] and in both, high-level β-galactosidase expression was observed in Schwann cells throughout the PNS.
Sequence analysis reveals a Krox-20 site that is neither essential nor sufficient for Schwann cell expression.a, Sequence analysis reveals multiple potential regulatory elements within SCE1, and only a few are shown. Among these is a Krox-20 site [(−)GCGTGGGTG; Sham et al., 1993]. This site lies near the 3′ end of SCE1, and consequently, was included in the 8.5 kb promoted construct that expressed only in oligodendrocytes, demonstrating that it is not sufficient for conferring Schwann cell expression. b, To determine whether the Krox-20 site is essential for SCE1-mediated expression, a construct bearing a mutation within the Krox-20 site was generated [(−)GCGTGGGTG→GCGGTTTCG, δKrox-20-SCE1]. The mutated SCE1 was ligated to the 3′ end oflacZ driven by 6 kb of MBP promoter.c, A P30 transgenic mouse bearing the δKrox-20-SCE1 construct demonstrates high-level reporter gene expression in both Schwann cells and oligodendrocytes, indicating that the Krox-20 site is not essential for SCE1 targeting function. Scale bar: c, 4.4 mm.
DISCUSSION
In this investigation, the expression of reporter constructs in transgenic mice was used to map DNA regulatory domains that controlMBP expression in Schwann cells. One robust Schwann cell enhancer was located and characterized, and we encountered evidence for additional enhancer activity in a further 5′ sequence. Our investigation also revealed activity of a developmentally regulated cell-specific repressor located in the more proximal MBPpromoter. Furthermore, SCE1, in combination with the heterologous hsp promoter, was found to drive expression in Schwann cells but not in oligodendrocytes, whereas MBP promoted constructs lacking the SCEs expressed only in oligodendrocytes. Despite the common role of MBP as a structural protein in both CNS and PNS myelin, the promoter organization we define here suggests that theMBP gene may be regulated, in large measure or entirely, through different transcription factors and regulatory elements in these two cell types.
The SCE sequences described here were found to direct reporter gene expression to Schwann cells in a position- and orientation-independent manner, consistent with their role as classic enhancers of transcription. Based on histochemical labeling intensity, SCE1 is maximally active during myelin elaboration and, like the endogenousMBP locus, SCE1-bearing reporter constructs are expressed at lower levels in mature mice. Although the developmental expression programs conferred by SCE1-containing constructs reflect major features of that reported for the endogenous MBPgene, no construct yet examined expresses in a manner that fully reflects normal MBP transcriptional output. Notably, differences in expression levels were encountered between sensory and motor fibers in mature mice in some SCE1-bearing lines, and some transgenes that were expressed robustly during myelination were shut off in mature animals. Consequently, additional Schwann cell regulatory sequence might remain to be located. Alternatively, a more normalized expression phenotype might be achieved by constructs in which contiguous SCE1 and SCE2 sequences are included.
Our investigations support a model in which the transcriptional regulation of MBP is achieved through multiple regulatory sequences sharing fundamental features with better characterized loci (Yuh et al., 1998). This organization would provide significant opportunity for the coordinately expressed myelin genes to share common regulatory components. A 1.1 kb 5′ flanking sequence from the P0 gene confers a program of expression to reporter constructs with similarities to that observed here for MBP SCE1 (Messing et al., 1992). In addition, 4 kb of 5′ flanking sequence from the CNPase gene drives expression in both oligodendrocytes and Schwann cells (Gravel et al., 1998). Consequently, it is plausible that elements within these promoters may bind identical transcription factors that are themselves regulated through shared signaling pathways. However, unless ordered and spaced similarly, the frequently short sequences that constitute functional transcription factor-binding sites would be difficult to identify by direct sequence comparisons. To date, no convincing sequence homology between SCE1 and the regulatory domains of other myelin genes has been identified.
Despite the clear role of axonal signals in the regulation of myelin genes, candidate molecules that could be implicated in the relevant signaling pathways are not abundant. Compounds and experimental conditions known to modulate myelin gene expression in Schwann cells and oligodendrocytes include progestins, which potentiate myelination both in vivo and in vitro, and electrical stimulation, which modulates the level of MBP expression in cultured Schwann cells (Koenig et al., 1995; Stevens et al., 1998). It also is well established that elevation of cAMP levels in cultured Schwann cells results in the upregulation of myelin gene expression (Lemke and Chao, 1988; Morgan et al., 1991). More recently, a role for myelin-associated glycoprotein (MAG) in bidirectional transduction of axon-Schwann cell signals has been proposed based on the observation that homozygous MAG null animals express both axonal and myelin anomalies (Li et al., 1994; Yin et al., 1998). However, a putative downstream effector of MAG, Fyn tyrosine kinase, does not appear to be required for PNS myelination (Fujita et al., 1998; Osterhout et al., 1999).
Within Schwann cells, one molecule with an essential role in myelinogenesis is the zinc finger transcription factor Krox-20. In mice bearing null Krox-20 alleles, Schwann cells make appropriate associations with axons, but subsequent myelin formation and the associated upregulation of myelin genes generally fails (Topilko et al., 1994). Krox-20 could effect its action by directly binding elements in the promoters of individual myelin genes, but our initial findings indicate that the one Krox-20 site recognized within SCE1 is neither sufficient nor necessary for Schwann cell targeting. Although this observation does not rule out a direct role for Krox-20 in modulating levels of myelin gene expression, it suggests that any targeting function mediated through Krox-20 would be achieved indirectly. The latter interpretation also is consistent with our inability to recognize Krox-20 binding sites within the 1.1 kb PO promoter, as well as the previous demonstration that during development, high-level Krox-20 expression precedes upregulation of myelin genes by several days (Murphy et al., 1996).
Last, the POU domain transcription factor SCIP has been implicated in the control of myelin gene expression. It is downregulated immediately before myelination, and both in vitro myelin gene promoter studies and a dominant negative transgenic model are consistent with it playing a role in repressing genes essential for Schwann cell maturation (Monuki et al., 1989, 1990; He et al., 1991; Weinstein et al., 1995). However, homozygous null SCIP mutants do myelinate, albeit in a delayed manner (Bermingham et al., 1996; Jaegle et al., 1996), leaving the precise role of SCIP unclear. Analysis of the SCE1 sequence did not reveal any SCIP sites.
Differential expression of SCE1-regulated constructs in those Schwann cells myelinating motor and sensory fibers of mature mice was unanticipated. Although this phenomenon was not detected in mice from lines in which histochemical labeling was particularly rapid and intense, it was a prominent feature of several moderately expressing lines. Therefore it is possible that significant differences may exist in all lines but be obscured in high-expressing lines. Differences recognized previously between myelinated motor and sensory fibers include the patterns of action potential trafficking and acquisition of the post-translational L2/HNK-1 epitope on several proteins in Schwann cells innervated by motor axons (Bowe et al., 1985; Martini et al., 1994). Because MBP is expressed constitutively by all myelin-bearing Schwann cells (Peterson and Bray, 1984), the striking difference in expression seen in motor and sensory fibers of mature mice from multiple SCE1 bearing transgenic lines appears to diverge from the normal expression pattern of the endogenous MBPlocus. This difference, presumably arising from differential use of elements within SCE1, suggests a novel level of Schwann cell heterogeneity directed by the innervating axon. Consequently, it is an intriguing possibility that the unequal susceptibility recognized for motor and sensory fiber types in certain human neuropathies and experimental animal models (Martini et al., 1995; Wrabetz et al., 2000) may arise at this level of Schwann cell heterogeneity.
Among the lines of transgenic mice established for this investigation, some consistently express the lacZ reporter in Schwann cells at exceptionally high levels and thus provide unique opportunities to introduce sensitive Schwann cell markers into culture, transplant, and regeneration preparations. In addition, some lines of mice bear reporter constructs containing limited numbers ofMBP regulatory elements, and these should provide novel opportunities to investigate the Schwann cell signaling pathways that control myelination (Farhadi et al., 1999). The robust expression observed in numerous lines bearing SCE1-regulated constructs suggests that SCE1 will provide a powerful tool for promoting gene expression in myelinating Schwann cells for both biological investigations and potential therapeutic interventions. Last, the 0.6 kb SCE1 defines a realistic target size within which the organization and function of the DNA regulatory elements of the MBPgene, along with their related transcription factors, can be investigated (Bronson et al., 1996; Forghani et al., 1999).
Footnotes
This work was supported by the Ludwig Institute (D.R.F.), the Canadian Neuroscience Network (L.G.), and the Medical Research Council of Canada (MRC) (R.F., H.F.F.). The Canadian Neuroscience Network, the Multiple Sclerosis Society of Canada, and the MRC supported parts of this investigation. We are grateful for the technical assistance provided by C. Artigas and J. Tremblay.
R.F. and L.G. contributed equally to this investigation.
Correspondence should be addressed to A. Peterson, Laboratory of Developmental Biology, Department of Neurology and Neurosurgery, Molecular Oncology Group H-5, McGill University, 687 Pine Avenue West, Montreal, Quebec, Canada, H3A 1A1. E-mail:alan{at}devbiol2.molonc.mcgill.ca.
L. Garofalo's present address: Health Canada, 1600 Scott Street, Ottawa, Ontario, Canada K1A 1B8.
D. R. Foran's present address: Department of Forensic Sciences, The George Washington University, 2036 H Street, North West, Washington, DC 20052.
P. Lepage's and T. Hudson's present address: Montreal Genome Centre, McGill University Health Center, 1650 Cedar Avenue, Room L3-401, Montreal, Quebec, Canada, H3G 1A4.