In the present study we provide data demonstrating that LC3, as a consequence of enhancing FN mRNA translation, increases CTGF levels, and this produces a faster growing, more adhesive and more invasive fibrosarcoma cell both in culture and in a living animal. Data from previous expression profiling of 239 soft tissue tumors showed that coordinated expression of both LC3 and CTGF was present in some invasive tumor types.
In our previous studies, we produced a variety of peptides from a recombinant LC3-GST fusion protein and determined that a 27bp region of LC3 appeared to be required for binding to the ARE of FN mRNA (our unpublished data). Present in this region is an RNA binding motif containing three consecutive arginine residues, that also influences mRNA translation (Lewis et al., 1998
) as well as mRNA stability (Shaw and Kamen, 1986
; Treisman, 1985
). However, prior to our observations, these arginine motifs (ARMs) were shown to have RNA binding properties only in proteins from bacteriophages and viruses (Weiss and Narayana, 1998
). As a control for our experiments, we constructed an LC3 transcript in which the ARM was mutated with an arginine to glutamine (LC3-R/Q) substitution that neutralized the positive charge, and that greatly reduced ARE binding activity. The mutant LC3 protein was stable and formed a doublet at 16kDa, but it appeared somewhat different from the WT protein in that the upper band was more intense than the lower band and it was also shifted slightly downward. Our previous studies related the lower band of the doublet to the phosphorylated form of LC3 that is associated with the pelleted membranes of the cell and with the polyribosomes necessary for mRNA translation (O’Blenes et al., 2001
). The lower band is also associated with a longer sequence (Kabeya et al., 2000
). The inability of LC3-R/Q to increase FN synthesis and thus maintain steady state levels of FN protein similar to those observed in LC3-WT transfected HT1080 cells is consistent with a relative impairment in translation of FN mRNA. This is supported by the polysome analysis that shows that in LC3-WT transfected cells, as opposed to LC3-R/Q mutant cells, there is an increase in FN mRNA transcripts in the heavy polysomes. Transfecting the LC3-WT cells with the rat WT FN construct, but not with an ARE-deleted mutant construct, also showed increased distribution of the WT FN rat mRNA in the heavy polysomes. This confirms previous studies by our group in vascular smooth muscle cells, that showed that ARE is critical to the function of LC3 in increasing FN mRNA translation (Zhou et al., 1997
). However, it is also clear from these and previous studies that the presence of an ARE alone does not confer increased translation in the presence of LC3. For example, the c-Myc transcript, has an ARE known to be an mRNA stability element, but LC3 does not regulate translation of c-Myc mRNA.
LC3 has functions in addition to translation of FN mRNA, and these include possible roles in microtubule assembly (Hammarback et al., 1991
), in mRNA transport (Seidenbecher et al., 2004
), and in autophagy (Kabeya et al., 2000
; Kabeya et al., 2004
). In the future, these functions could be assessed in the HT1080 cells transfected with LC3. Interestingly, loss of function of LC3 in a knockout mouse reported by our group, has no autophagy phenotype (Cann et al., 2008
The LC3 transfected HT1080 cells appeared to be more highly proliferative both in culture and when implanted into SCID mice. Studies in cultured cells were subsequently carried out to determine whether we could account, at least in part, for this phenotype by an LC3-mediated increase in FN synthesis alone. Previous studies have attributed a FN-mediated increase in the adhesive properties of tumor cells as being responsible for their reduced proliferation (Dean et al., 1988
). More recent studies have shown that fibronectin interaction with α5
(Aguirre-Ghiso et al., 2003
) as well as α3
integrins increases tumor cell growth, as well as invasion by activating MMP-9 and Rac1 (Wei et al., 2007
). Hence, the context is clearly important, and suggests that in response to LC3, coordinate regulation of genes at the transcriptional and post-transcriptional level may be necessary to produce the FN-dependent proliferative and invasive responses. In keeping with this, our studies show that LC3, via FN, also promotes cell adhesion, a property previously shown to be necessary for the migration of fibrosarcoma cells (Zaman et al., 2006
) and vascular cells (Boudreau et al., 1991
) in 3D matrices.
To address whether FN-cell interaction might be inducing other genes that are required to enhance HT1080 cell adhesion, invasion and proliferation, we carried out microarray analysis to compare LC3-WT and vector transfected HT1080 cells. Although a number of transcripts were upregulated, one, CTGF, stood out as being critical to the mechanism associated with our findings, and its upregulation was confirmed by qRT-PCR to be both LC3 and FN-dependent. While it is possible that LC3 increases CTGF mRNA by increasing mRNA stability, the dependence of enhanced CTGF expression on FN, suggests an indirect effect of LC3. Overexpression of CTGF has been shown in a number of cancers (Croci et al., 2004
; Koliopanos et al., 2002
; Kubo et al., 1998
; Moritani et al., 2003
; Pan et al., 2002
; Shakunaga et al., 2000
; Vorwerk et al., 2002
; Wenger et al., 1999
; Xie et al., 2001
; Xie et al., 2004
; Zeng et al., 2004
), but its direct role in tumor suppression or progression has not been characterized.
CTGF is a cysteine-rich secreted protein, belonging to a group of immediate-early genes induced by growth factors, such as TGF-β, or certain oncogenes. CTGF promotes proliferation and migration of vascular endothelial cells (Takigawa et al., 2003
) and stimulates human mesangial cell adhesion to fibronectin (Weston et al., 2003
). CTGF stimulates mesenchymal cells, including fibroblasts, chondrocytes and osteoblasts, to proliferate and to produce connective tissue components such as collagen type 1 and FN, while also remodeling the extracellular matrix (Blom et al., 2001
; Frazier et al., 1996
), and promoting granulation tissue formation (Chen et al., 2001
). In mice, CTGF mediates endothelial cell adhesion and migration through interaction with the αv
integrin, and induces endothelial cell survival and angiogenesis (Shimo et al., 2001
). Thus, heightened expression of CTGF may contribute to the angiogenic response that supports HT1080 cell growth in addition to the proliferative and invasive response of the cells per-se.
It is interesting that whereas CTGF was shown to induce FN (Blom et al., 2001
; Frazier et al., 1996
), in our study, the expression of CTGF is FN-dependent. This may imply a positive feedback mechanism that promotes the features of tumorigenesis by amplifying the interaction of both FN and CTGF with integrin receptors, including αν
(Chen et al., 2001
; Gao and Brigstock, 2006
; Jedsadayanmata et al., 1999
; Lymn et al., 2002
). Future studies could address whether loss of CTGF also reduces FN. In addition, it would be of interest to further investigate whether LC3-mediated enhanced synthesis of FN can, by inducing integrin linked kinase (ILK), mediate transcriptional activity of factors upstream of the CTGF promoter, such as AP1 (Troussard et al., 2000
; Troussard et al., 1999
). ILK activation downstream of fibronectin-integrin interaction can induce an invasive phenotype via AP-1-dependent upregulation of matrix metalloproteinase MMP-9 (Troussard et al., 2000
). It is intriguing, however, that FN induced production of CTGF was sufficient to explain the proliferative and invasive phenotype of the cultured HT1080 cells.
Our subsequent studies show co-expression of LC3 and CTGF mRNA in at least some soft tissue tumor types such as desmoid type fibromatosis but not others. Further prospective phenotyping studies and confirmation of the microarray data by qRT-PCR are necessary to indicate whether these factors could serve as important new biomarkers. Our previous studies have shown that TNF alpha upregulates FN in coronary artery smooth muscle cells via LC3 (O’Blenes et al., 2001
) and TNF alpha is known to upregulate CTGF (Cooker et al., 2007
), so it would be interesting to know whether TNF could mediate an increase in CTGF and FN via LC3 in HT1080 cells. TGF beta also stimulates production of FN (Dean et al., 1988
). While there is no obvious coordinate expression of these cytokines with CTGF and LC3, it would be interesting to establish in future prospective studies, whether the subset of tumors with the increase in LC3 and CTGF have a cytokine signature.