Some time ago, it was discovered that the combinatorial expression of LIM homeodomain (LIM-HD) transcription factors defines subclasses of motor neurons that segregate into columns in the spinal cord (Tsuchida et al.,1994). Indeed, LIM-HD genes have been identified as crucial for the development of specific cell populations in the spinal cord such as motor- and interneurons (Shirasaki and Pfaff,2002; Thaler et al.,2002), and for the formation of many other neuronal and non-neuronal structures (Bach,2000; Hobert and Westphal,2000). The identification of the LIM domain as a protein-protein interaction domain (Schmeichel and Beckerle,1994) resulted in the isolation of several LIM domain-interacting cofactors (Bach,2000; Matthews and Visvader,2003), including the homodimer-forming CLIM cofactor family that consists of CLIM1/Ldb2 and CLIM2/NLI/Ldb1/Chip (Jurata et al.,1996,1998; Agulnick et al.,1996; Bach et al.,1997; Morcillo et al.,1997) and the RING finger protein RLIM (Bach et al.,1999). A requirement of CLIM cofactors for at least some of the in vivo activities of LIM-HD proteins has been demonstrated. Indeed, the association of CLIM cofactors to LIM-HD proteins has been shown to be crucial for the development of specific types of interneurons and motorneurons during neural tube development (Thaler et al.,2002). In addition, it has been shown that cellular CLIM protein concentrations are critical determinants of LIM-HD activity in vivo (Matthews and Visvader,2003). RLIM has been identified as a ubiquitin ligase, able to target CLIM for proteasomal degradation, thereby inhibiting developmental LIM-HD activity (Ostendorff et al.,2002; Hiratani et al.,2003). Although the expression pattern of LIM-HD proteins has been analyzed in detail (Hobert and Westphal,2000), little is known about the expression of LIM cofactors at the protein level. Here, we have compared the expression of RLIM and CLIM during mouse neural tube development. We demonstrate that both cofactors display similar expression domains in the ventral neural tube where LIM-HD proteins are expressed. We show further that CLIM and RLIM expression is widely detected from E8.5 to E10.5, then becomes restricted to ventral expression domains at E11.5–E13, and subsequently weakens to low levels around E14.5. As concentrations of LIM cofactors have been found critical for LIM-HD function, our results indicate developmental regulation of LIM-HD activity at the cofactor level.
RESULTS AND DISCUSSION
The crucial influence of LIM cofactors CLIM and RLIM on the in vivo activity of LIM-HD proteins has been demonstrated (Bach,2000; Matthews and Visvader,2003). At the mRNA level, it has previously been shown that both types of cofactors are expressed early during mouse embryogenesis in a ubiquitous manner with higher levels in tissues expressing LIM-HD factors (Bach et al.,1997,1999). To investigate the developmental LIM cofactor expression at the protein level in more detail, we have raised specific antisera in rabbits and guinea-pigs. In Western blots, both the RLIM and CLIM antisera recognized their antigen and showed no detectable cross-reaction in protein extracts prepared from the pituitary αT3 cell line (Fig. 1), which express relatively high levels of both proteins (Ostendorff et al.,2002). The Western blot analysis further revealed that the antiserum directed against the CLIM protein recognized both CLIM1 and CLIM2 in these experiments. This result, combined with the fact that both proteins are highly conserved, strongly argues that the polyclonal CLIM antiserum also recognized both proteins in the following immunohistochemical studies. In immunohistochemical experiments, we confirmed the specificity of antisera directed against CLIM and RLIM by competition experiments, adding excess amounts of the respective antigen (data not shown).
After having demonstrated the high specificity of our antisera, we performed immunohistochemistry to investigate and compare the developmental expression profile of each cofactor on whole mouse embryos at developmental stages E8.5 and E9.5 (Fig. 2). At these stages, we detected widespread expression of RLIM (Fig. 2A,C,D) and CLIM (Fig. 2F,H,I), with higher expression in the developing neural tube. Both RLIM and CLIM were highly expressed in dorsal areas of the neural tube before and after closure. Because of the known functions of LIM-HD proteins during neural tube development (Shirasaki and Pfaff,2002), we focused our further analyses on this structure. By performing immunohistochemistry on embryonic sections, we observed widespread expression of cofactors RLIM and CLIM with higher expression in dorsal regions of the developing neural tube at E8.5 and 9.5 (Fig. 2), confirming the results obtained on whole embryos.
We used vibratome sections on mouse embryos to elucidate the developmental expression profile of LIM cofactors from E11.5 to E15.5 at the thoracic level. These sections are of around 50–80 μm thickness corresponding to several cell diameters, allowing to better visualize regions with high protein expression. At E11.5, we observed higher expression of both RLIM and CLIM in dorsal and ventral neural tube expression domains (Fig. 3). Expression of both proteins was not only detected in ventral neural tube areas where post-mitotic neurons are localized but also in regions of the developing neural tube that contain progenitor neurons. CLIM expression in progenitor regions was localized in the dorsal half of the neural tube, whereas RLIM expression in progenitor regions extended also in more ventral regions. The expression of RLIM and CLIM condensed at E12.5 until at E13.5, when dorsal expression faded and both cofactors are detected mainly in ventral neural tube regions. Interestingly, already at E12.5 CLIM expression in dorsal regions seemed weaker when compared to RLIM, suggesting that the fading of CLIM proteins in dorsal regions precedes that of RLIM. At E14.5, signal strengths of ventral expression fades until at E15.5 only relatively little expression of RLIM and CLIM is detected at the protein level (Fig. 3, and data not shown). Although it is difficult to assess to what extent the down-regulation of RLIM and CLIM expression occurs, it seems clear that the regionalized expression seen at E11.5–12.5 dramatically changes at later stages of mouse neural tube development. This down-regulation was surprising as mRNAs of CLIM1,2 and RLIM are expressed at these stages (Bach et al.,1999), suggesting that post-translational regulation occurs. Therefore, we next investigated the mRNA expression of LIM cofactors during mouse development. Northern blot analysis using specific probes for RLIM, CLIM1, and CLIM2 on a blot containing RNA prepared from whole mouse embryos at different embryonic stages revealed that the mRNAs encoding these proteins were synthesized throughout all developmental stages analyzed (Fig. 4A). This result is in agreement with our previous studies that showed widespread expression of mRNAs encoding these cofactors (Bach et al.,1999). Indeed, we found widespread expression in the neural tube as well as DRG (blue arrows) of both RLIM and CLIM2 mRNA at E12.5 and E14.5. At E12.5 CLIM1 expression was more confined to ventral neural tube regions (Fig. 4B). This mRNA expression indicates an important contribution of CLIM1 to the strong protein expression detected in this region at this time-point (Fig. 3), whereas the contribution of CLIM2 in the ventral neural tube remains unclear. However, since there is little CLIM1 mRNA expression in dorsal neural tube areas at E12–12.5, it is likely that the CLIM staining at the protein level observed in these areas reflects CLIM2 expression. At E14.5, we detected CLIM1 mRNA also in the dorsal neural tube, as well as strong staining in DRG tissue. These experiments show that CLIM1 mRNA is expressed in specific embryonic regions including neural tube and DRG in a dynamic fashion during mouse embryogenesis, in contrast to CLIM2 and RLIM mRNAs, which are widely distributed. As mRNA and protein expression of RLIM and also CLIM clearly differ, our results suggest additional post-translational mechanisms regulating LIM cofactor expression/stability.
To further define the expression domains of RLIM and CLIM during neural tube development, we performed co-immunohistochemistry on thoracic sections of E12.5 embryos, co-staining with dorsal and ventral marker proteins. Investigating the ventral border of the dorsal expression of LIM cofactors in the developing neural tube we performed co-stainings using the Pax7 antibody (Ericson et al.,1996). We found that the ventral border of Pax7 expression co-incited with expression of RLIM and CLIM (Fig. 5A,B). LIM-HD expression in this region of the developing neural tube is absent, so it is interesting to note that CLIM cofactors have been associated with functions of other classes of homeodomain proteins (Bach et al.,1997; Torigoi et al.,2000). We next investigated the ventral neural tube expression of both cofactors by performing co-immunohistochemistry using antisera directed against LIM-HD proteins Isl1,2 and Lhx3. Indeed, co-stainings using the RLIM and Isl1,2 antisera demonstrated that both classes of proteins co-localize in ventral neurons with RLIM displaying a wider distribution than Isl1,2 proteins (Fig. 5C). In neighboring dorsal root ganglia (DRG), we also detected cell populations in which high levels of RLIM co-localized with Isl proteins. Interestingly, we detected cell populations that expressed high relative levels of Isl proteins and low levels of RLIM and vice versa (Fig. 5C). We also found co-localization of CLIM cofactors with Isl1,2 proteins in ventral neural tube neurons as previously reported (Jurata et al.,1996; Thaler et al.,2002) and in subpopulations of DRG cells (Fig. 5D). Co-stainings using a rabbit antiserum directed against Lhx3 and our guinea pig RLIM antiserum show that all Lhx3-positive cells in the ventral neural tube also expressed high levels of RLIM (Fig. 5E). These results demonstrate that high levels of RLIM and CLIM cofactors co-localize with LIM-HD proteins Isl1,2 and Lhx3. However, using an antibody that recognizes lim-1,2, we found that high levels of lim1,2 neither co-localized with RLIM nor with CLIM cofactors (data not shown), suggesting possible functions of LIM-HD proteins without cofactors. Because of the surprisingly similar expression patterns of RLIM and CLIM, we tested whether both proteins are co-localized. The co-stainings with RLIM and CLIM antisera revealed a striking co-localization in ventral neural tube neurons and DRG neurons (Fig. 5F). Although the amounts of both proteins appear to differ in various cells, this result is rather unexpected as CLIM cofactors are targeted for proteasomal degradation by the ubiquitin ligase RLIM.
The high relative protein levels of RLIM in ventral neural tube regions that do not express particularly high RLIM mRNA levels indicate that RLIM protein is more stable in cells located in this region. Since RLIM is a RING finger ubiquitin ligase able to autoubiquitinate and target itself for proteasomal degradation (Ostendorff et al.,2002), these results suggest that RLIM's ubiquitin ligase activity may be locally regulated. The high levels of RLIM co-localized with its substrate protein CLIM in ventral neural tube neurons is a further indication that RLIM's role to target CLIM for proteasomal degradation is not exerted in all cells at all times. Thus, our observations open up the possibility that RLIM's activity to target CLIM for degradation may be locally regulated. In this context, it is interesting to note that it has been reported in specific cases that the activity of ubiquitin ligases can be regulated, often by phosphorylation events (Pickart,2001; Bach and Ostendorff,2003). Further research is needed to address this question in more detail.
In summary, our results show that LIM cofactors RLIM and CLIM are dynamically expressed in an overlapping fashion during neural tube development with a strong predicted impact on local LIM-HD activity.
Antibodies and Immunohistochemistry
Polyclonal RLIM antisera raised in guinea pig was carried out as described previously for rabbit (Ostendorff et al.,2002) against the same domain. The CLIM and RLIM rabbit antisera have been described earlier (Ostendorff et al.,2002). Islet-1 40.2D6, lim-1, and Pax-7 monoclonal antibodies were purchased from Developmental Studies Hybridoma Bank, Iowa City, IA. We performed the immunohistochemical experiments on whole-mounted embryos and on cryosections as previously described (Tursun et al.,2005, Briscoe et al.,2000). Biotinylated secondary antibodies for the immunohistochemistry were goat anti-rabbit and goat anti-guinea pig (Vector). For immunohistochemistry, we used Alexa 488 goat anti-mouse (MoBiTec), Alexa 488 goat anti-guinea pig (MoBiTec), Cy5 goat anti-rabbit (MoBiTec), and Cy5 goat anti-rat (MoBiTec) as secondary antibodies.
In Situ Hybridizations and Northern Blot Analysis
In situ hybridizations with specific CLIM and RLIM probes were carried out essentially as described (Bach et al.,1997,1999) with the modification that we used Dab vector kit according to the manufacturer's instruction. Northern blot analysis on RNA prepared from whole embryos (Clontech) was carried out as described (Bach et al.,1991) using specific 32P-labeled probes (Bach et al.,1999). The same blot was hybridized consecutively with probes for RLIM, CLIM1, CLIM2, and β-actin as loading control.
H.P.O. and B.T. were supported by an EMBO short-term fellowship and a doctoral fellowship from the Boehringer Ingelheim Foundation, respectively. I.B. was supported by a Heisenberg scholarship of the Deutsche Forschungsgemeinschaft (DFG) and the Chica and Heinz Schaller foundation. This work was supported by grants from the DFG to I.B.