MIG-10 (Lamellipodin) stabilizes invading cell adhesion to basement membrane and is a negative transcriptional target of EGL-43 in C. elegans

Abstract

Cell invasion through basement membrane (BM) occurs in many physiological and pathological contexts. MIG-10, the C. elegans Lamellipodin (Lpd), regulates diverse biological processes. Its function and regulation in cell invasive behavior remain unclear. Using anchor cell (AC) invasion in C. elegans as an in vivo invasion model, we have previously found that mig-10’s activity is largely outside of UNC-6 (netrin) signaling, a chemical cue directing AC invasion. We have shown that MIG-10 is a target of the transcription factor FOS-1A and facilitates BM breaching. Combining genetics and imaging analyses, we report that MIG-10 synergizes with UNC-6 to promote AC attachment to the BM, revealing a functional role for MIG-10 in stabilizing AC-BM adhesion. MIG-10 is also required for F-actin accumulation in the absence of UNC-6. Further, we identify mig-10 as a transcriptional target negatively regulated by EGL-43A (C. elegans Evi-1 proto-oncogene), a transcription factor positively controlled by FOS-1A. The revelation of this negative regulation unmasks an incoherent feedforward circuit existing among fos-1, egl-43 and mig-10. Moreover, our study suggests the functional importance of the negative regulation on mig-10 expression by showing that excessive MIG-10 impairs AC invasion. Thus, we provide new insight into MIG-10’s function and its complex transcriptional regulation during cell invasive behavior.

1. Introduction

Cell invasion takes place in diverse contexts, including organogenesis, morphogenesis, immune surveillance, wound healing, and regeneration [1]. A common barrier the invading cells have to navigate through is basement membrane (BM), a dense, 50–100 nm thick, sheet-like structure consisting of highly-crosslinked and self-assembled glycoproteins and proteoglycans [2]. Defects in the cell invasive machinery contribute to various human diseases, including asthma, pre-eclampia [3,4], and importantly, cancer metastasis, which accounts for 90% mortality in solid cancers [5]. Thus, understanding how cells invade through BM in physiological contexts could help identify novel targets for treating these diseases.

Cell invasion through BM requires the coordination of multiple interlinked steps beginning with establishing direct contact between cells and their invading target, BM. During this establishment, diverse cellular processes, such as cytoskeleton reorganization [6] and gene transcription [7,8], are precisely regulated towards successful BM breaching.

Anchor cell (AC) invasion in C. elegans has been developed to be an in vivo invasion model (Fig. 1A) [7,9,10]. During the L2/L3 larval stage of C. elegans hermaphrodite development, the AC, a highly specialized gonadal cell, invades through the gonadal and vulval BMs and contacts the central primary vulval precursor cells (1° VPC). This invasion initiates the connection between the uterine and vulval cells [10]. Prior to invasion (1° VPC P6.p 1-cell stage), UNC-6 (netrin) secreted from the ventral nerve cord (VNC) and integrin signaling from within the AC synergistically promote the formation of a specialized F-actin-rich invasive membrane domain with polarized actin regulators at the basal membrane of the AC in contact with the BM [9]. Approximately five hours later, when the P6.p cell gives rise to two daughter cells (2-cell stage), the AC initiates invasion by generating invasion protrusions. The transcription factor FOS-1A is required for breaching and further removing BM [7]. AC invasion is completed by the time the P6.p descendants divide again, generating four granddaughter cells (4-cell stage). At this time, the AC creates a gap in the BM the size of the AC.

Fig. 1. Schematics of AC invasion and localization of MIG-10B in unc-6 mutants.

Anterior is left; ventral is down; and arrows point to the AC in this and all other figures. (A) A schematic diagram illustrates AC invasion in C. elegans. In the early L3 larva the AC is attached to the basement membrane (BM, light blue) over the primary vulval precursor cell (1° VPC) (light green, P6.p 1-cell stage, left). At this time UNC-6 (netrin) (blue arrows) secreted from the ventral nerve cord (VNC) polarizes its receptor UNC-40 (blue ovals) and F-actin (orange) to the invasive cell membrane in contact with BM in the presence of integrin (magenta). During the mid-L3 stage, after the P6.p cell divides (P6.p 2-cell stage, middle), the AC breaches the BM and generates an invasive protrusion that invades between the two central 1° VPC granddaughter cells by the late L3 (P6.p 4-cell stage, right). The transcription factor FOS-1A promotes BM breaching. UNC-40 (DCC) mediates protrusion formation. (B and C) DIC images (left) and corresponding fluorescence (right). MIG-10B was similarly polarized at the AC’s basal membrane (arrowheads) in the wild-type and unc-6 mutants at the P6.p 2-cell stage. (D) Quantification of MIG-10B polarization at the P6.p 1- 2-cell and 4-cell stages in wild-type animals and unc-6 mutants (n ≥ 10 per stage per genotype, Student’s t-test). In this and all other figures, *, p < 0.05; **, p < 0.01; ***, p < 0.001; n.s., not significant. Error bars, standard error of mean. Scale, 5 μm.

C. elegans MIG-10 is a member of the MIG-10/RIAM/Lamellipodin (MRL) adaptor protein family that plays an important role in cell migration [11]. MIG-10 regulates axon guidance, neuronal migration [12], the clustering of vesicles at the synapse [13], and the outgrowth of the processes of the excretory cell [14]. In AC invasion, we have previously found that mig-10 functions largely outside of unc-6 (netrin) signaling [15] in contrast to mig-10 acting downstream of unc-6 in axon guidance [12]. While MIG-10 polarization to the invasive membrane is dependent on the extracellular matrix receptor integrin heterodimers (INA-1/PAT-3), mig-10 expression is positively controlled by FOS-1A, which regulates the expression of genes that promote BM breaching [15]. Given that mig-10 acts redundantly with other FOS-1A target genes, mig-10 was thought to help breach BM [15].

Here we report that MIG-10 synergizes with UNC-6 to promote AC attachment to the BM in a cell-autonomous manner prior to and throughout invasion. MIG-10 regulates F-actin accumulation in the absence of UNC-6. Moreover, EGL-43, a transcription factor downstream of FOS-1A, negatively regulates mig-10 expression, which together with the known regulatory relationships between fos-1 and egl-43/mig-10 forms an incoherent (type I) feedforward transcriptional loop. Further, we suggest that the negative regulation on mig-10 expression is functionally important as overexpressed MIG-10 disrupts AC invasion. Thus, we identify a functional role for MIG-10 in promoting cell-matrix adhesion in AC invasion and unveil a new layer of the complicated transcriptional regulation on mig-10.

2. Materials and methods

2.1 Worm handling and strains

Worms were reared under standard conditions at 15°C, 20°C, or 25°C [16]. N2 Bristol strain was used as wild-type. We use a “>“ symbol for linkages to a promoter and a “::” symbol for linkages that fuse open reading frames. The following alleles and transgenes were used: qyIs57[cdh-3>mCherry::moeABD], qyIs183[cdh-3>GFP::mig-10b;cdh-3>mCherry], sIs14214[mig-10b>GFP], cuwEx1[cdh-3>GFP::mig-10b (overexpressed; higher level); cdh-3>mCherry::moeABD]; cuwEx2[cdh-3>GFP::mig-10b (overexpressed; lower level); cdh-3>mCherry::moeABD]. Linkage Group: LGII, rrf-3(pk1426); LGIII, mig-10(ct41); LGX, unc-6(ev400).

2.2 Microscopy, image acquisition, and image processing

Images were acquired using an Olympus BX53 microscope with a 100x UPlanSApo objective (NA 1.4) and a XM10 CCD camera controlled by cellSens Entry software (Olympus, Japan), or a Zeiss AxioImager microscope with a 100x Plan-APOCHROMAT objective (NA 1.4) and a Zeiss AxioCam MRm CCD camera controlled by Axiovision software (Zeiss Microimaging), or Zeiss AxioImager microscope with a 100x Plan-APOCHROMAT objective (NA 1.4) and a Yokogawa CSU-10 spinning disc confocal controlled by iVision software (Biovision Technologies). Acquired images were processed using ImageJ 1.40 and Photoshop CS5 (Adobe Systems). 3D reconstructions were generated from confocal z-stacks, analyzed and exported using Imaris 7.4 (Bitplane). Graphs and figures were made using Adobe Illustrator CS5 (Adobe Systems).

2.3 Quantitative measurements of MIG-10B and F-actin polarization and the expression level of mig-10b transcriptional reporter

MIG-10B and F-actin polarization was determined as the percentage of the fluorescence intensity at the invasive membrane region over the total fluorescence intensity within the entire cell (nucleus excluded) using ImageJ 1.40. EGL-43 positively controls cdh-3 promoter [17]. For quantifying F-actin polarity after egl-43a RNAi, the ACs that failed to invade and had mCherry::moeABD expression at the reasonable level for imaging were used for analysis. For quantifying mig-10b>GFP expression, the total fluorescence intensity within the ACs in sIs14214[mig-10b>GFP] animals that were treated with either L4440 control RNAi or egl-43a RNAi, was measured using ImageJ 1.40.

2.4 RNA interference

Double-stranded-RNA(dsRNA)-mediated gene interference was conducted by feeding larvae with bacteria expressing dsRNA using standard techniques [18]. To bypass egl-43a RNAi’s early effect on AC/VU (anchor cell/ventral uterine) fate specification, synchronized L1-arrested larvae of qyIs57, sIs14214[mig-10b>GFP], rrf-3(pk1426), and rrf-3(pk1426);mig-10(ct41) were grown for 2 hours on regular OP50 bacteria at 20°C before transferred to egl-43a RNAi plates.

2.5 Quantitative analysis of F-actin volume

F-actin volume was measured as previously described [19]. Briefly, confocal z-stacks of F-actin networks in the AC expressing the F-actin binding probe (mCherry::moeABD) were collected. 3D reconstructions of F-actin were built using these z-stacks in Imaris 7.4 (Bitplane). Isosurface renderings of mCherry::moeABD were created using Imaris “isosurface rendering” function by setting a threshold that outlined the dense F-actin network. Quantitative measurements were made for the volume of fluorescent intensity within these isosurface renderings [19].

2.6 Molecular biology and transgenic strains

The RNAi clone specifically targeting egl-43a was generated by PCR amplifying the unique coding region (1^st bp ~ 636^th bp) of egl-43a cDNA. This fragment was inserted into the L4440 plasmid at XbaI and HindIII. Standard techniques were used to generate transgenic animals [20]. The constructs cdh-3>GFP::mig-10b [15] and cdh-3>mCherry::moeABD [9] were used to generate cuwEx1 and cuwEx2 transgenic animals (Table S1). Transgenic worms were created by co-injecting expression constructs with the transformation marker pPD#MM016B (unc-119+) and the co-injection marker (myo-2>GFP) into the germline of unc-119(ed4) mutants. These markers were injected with EcoRI-digested salmon sperm DNA and pBluescript II at 50 ng/μl as carrier DNA along with the expression constructs.

2.7 Statistics

Statistical analyses were performed using Student’s t-test or Fisher’s exact test as indicated in the text.

3. Results and discussion

3.1 MIG-10B remains polarized at the invasive membrane of the AC in unc-6 mutants

Among mig-10’s three different isoforms, mig-10b is the only isoform specifically expressed in the AC [15]. To further study MIG-10B’s function in AC invasion, we first examined MIG-10B’s localization in unc-6 mutant ACs. Mig-10b expression was driven by the AC-specific promoter cdh-3 (qyIs183[cdh-3>GFP::mig-10b]). At this expression level, AC invasion was not disturbed (n=20/20 invaded at the P6.p 4-cell stage). In wild-type animals, approximately 50~55% of the total amount of MIG-10B was polarized at the AC’s basal plasma membrane prior to and during AC invasion (Fig. 1B and D). MIG-10B was similarly polarized in unc-6 mutants (Fig. 1C and D), indicating that MIG-10B polarization is independent of UNC-6, consistent with the majority of MIG-10B’s activity outside of netrin signaling [15].

3.2 MIG-10B synergizes with UNC-6 to promote AC adhesion to the BM

Loss of mig-10 alone causes mild invasion defects [15], suggesting that mig-10 acts with other genes during invasion. We next examined AC invasion in the double null mutants unc-6(ev400);mig-10(ct41). In unc-6(ev400) single mutants, all ACs attached to the BM at the P6.p 1-cell stage prior to the invasion (Fig. 2A and C). The ACs later started to detach from the BM staying afloat in the gonad. The percentage of detached ACs increased to 10% and 15% at the P6.p 2- and 4-cell stages, respectively (Fig. 2C), suggesting that UNC-6 promotes AC-BM adhesion during AC invasion. In contrast, mig-10(ct41) mutants had no phenotypic defects in AC-BM attachment. However, when mig-10 was removed in unc-6 mutants, AC detachment was observed at the P6.p 1-cell stage, revealing an early requirement for MIG-10B in stabilizing AC-BM attachment (Fig. 2B and C). AC detachment was later drastically increased by approximately 4–6 folds at the P6.p 2- and 4-cell stages (Fig. 2C), indicating a synergistic effect between MIG-10B and UNC-6 on mediating AC-BM adhesion during AC invasion. To determine MIG-10B’s site of action in adhesion regulation, we expressed mig-10b under the AC-specific promoter cdh-3 in unc-6;mig-10 doubles. AC-specific expression of mig-10b rescued detachment and restored AC-BM association to the level similar to unc-6 single mutants (Fig. 2C), suggesting a cell-autonomous role for MIG-10B in regulating adhesion. Together, these results reveal a progressively increased requirement for MIG-10B within the AC for AC-BM association in the absence of UNC-6.

Fig. 2. Loss of mig-10 enhances AC detachment from the BM and reduces F-actin accumulation in the absence of UNC-6.

(A and B) DIC images (left), corresponding fluorescence (middle), and overlay (right). (A) In unc-6 mutants the AC (visualized by F-actin binding probe cdh-3> mCherry::moeABD) failed to invade and attached to the intact BM (arrowheads; orange dotted line). (B) In unc-6;mig-10 double mutants, the AC failed to invade and detached from the BM. (C) Quantification of AC detachment percentage in unc-6 and unc-6;mig-10 mutants at the P6.p 1-, 2- and 4-cell stages, and in unc-6;mig-10;qyIs183 at the P6.p 4-cell stage. The total numbers of animals scored and significant differences are indicated (Fisher’s exact test). (D and E) 3D reconstructions of confocal z-stacks from the ACs at the P6.p 4-cell stage. Fluorescence (left), overlay of fluorescence and corresponding F-actin networks rendered with isosurfaces (right). (D) In unc-6 mutants F-actin (visualized by F-actin probe cdh-3>mCherry::moeABD) was mislocalized to the AC’s apical-lateral membranes (arrowheads). (E) In unc-6;mig-10 doubles F-actin remained mispolarized. The volume of F-actin was significantly reduced. (F) Quantification of the normalized total volume of F-actin in the wild-type (normalization control), unc-6, mig-10, unc-6;mig-10 mutants at the P6.p 2- and 4-cell stages (n ≥ 15 per stage per genotype). Significant differences are indicated (Student’s t-test).

Given that MIG-10B’s polarized localization at the invasive membrane depends on integrin [15], MIG-10B may function within integrin signaling to regulate cell-matrix association. Interestingly, RIAM and Lamellipodin (Lpd), another two members of the MRL protein family, are able to activate integrin [21,22,23], but elicit opposing effects on adhesion [21]. Under simplified in vitro cell culture conditions, RIAM promotes cell adhesion, whereas Lpd reduces adhesion. Although it remains to be determined whether RIAM and Lpd also exert differential regulatory effects on adhesion in vivo, our results provide evidence supporting MIG-10B’s in vivo role in promoting cell-matrix adhesion.

3.3 MIG-10B promotes F-actin accumulation in unc-6 mutants

Cell-matrix adhesion is tightly coupled with actin dynamics regulation [24]. To determine how loss of mig-10 enhanced AC detachment from the BM, we examined MIG-10B’s effect on F-actin in unc-6 mutants. Consistent with the previous report [19], loss of unc-6 alone disrupted only F-actin polarity without affecting F-actin total amount in comparison to the wild-type (Fig. 2D and F; Fig. S1). While loss of mig-10 in unc-6 mutants did not further disrupt F-actin polarity (Fig. S1), it did reduce F-actin amount by 68% and 67% at the P6.p 2- and 4-cell stages, respectively (Fig. 2D, E and F), suggesting that MIG-10B promotes F-actin accumulation in unc-6 mutant ACs, consistent with the MRL family proteins’ role in regulating actin cytoskeleton [15]. Given this synergistic effect between MIG-10B and UNC-6, it is possible that MIG-10B is involved in regulating actin regulators downstream of UNC-6, such as unc-34(Ena/VASP), mig-2/Rac, ced-10/Rac [15], or other unidentified players. More importantly, this marked reduction in F-actin amount occurred concurrently with a significant enhancement in AC detachment, raising a possibility that MIG-10B mediates AC-BM adhesion through maintaining F-actin accumulation in the absence of UNC-6.

3.4 EGL-43A suppresses mig-10b expression and forms an incoherent (type I) feedforward loop (IFFL) with FOS-1A and MIG-10B

Although the synergistic effects between MIG-10B and UNC-6 suggest the downstream convergence that warrants future investigation, these effects also indicate that MIG-10B has the activity controlled by netrin-independent pathways. FOS-1A, a key transcription factor during AC invasion, controls a complicated gene regulatory network involving mig-10 [15] and egl-43 (a transcription factor [25]). To further investigate the interactions between mig-10b and the fos-1 network components, we examined egl-43’s effect on mig-10b expression using a reporter containing 5′ cis-regulatory elements (2.9 kb sequence immediately upstream of mig-10b start codon) to drive GFP [15]. Egl-43, the C. elegans Evi-1 proto-oncogene, encodes two different transcripts. The longer isoform egl-43a is transcriptionally controlled by FOS-1A and required for AC invasion, whereas the shorter isoform egl-43b is not regulated by FOS-1A [17]. The role of egl-43b in AC invasion is unclear because egl-43b’s open reading frame with 3′ UTR is contained in egl-43a (Fig. S2), making it difficult to downregulate egl-43b without affecting egl-43a [17]. We thus specifically targeted egl-43a with an RNAi clone containing the sequence against the unique coding region of egl-43a (Fig. S2). The reduction in EGL-43A blocked AC invasion (Fig. 3A and B), consistent with the previous report [17]. Intriguingly, this reduction increased mig-10b expression prior to (approximately 2.5 folds for the p6.p 1-cell stage) and throughout AC invasion (2 folds for the p6.p 2- and 4-cell stages) (Fig. 3C). Moreover, while wild-type animals had mig-10b expression peaked at the P6.p 2-cell stage, egl-43a RNAi resulted in the early appearance of this peak at the P6.p 1-cell stage (Fig. 3C). Together, these results indicate that EGL-43A suppresses mig-10b expression and delays peaking of mig-10b expression.

Fig. 3. EGL-43A negatively regulates mig-10b expression and forms an incoherent (type I) feedforward loop with FOS-1A and MIG-10B.

(A and B) DIC images (left), corresponding fluorescence (right). (A) In the animal treated with L4440 control RNAi, the AC invaded normally at the P6.p 4-cell stage (arrowhead) and mig-10b (mig-10b>GFP) expression was detected in the AC. (B) RNAi targeting egl-43a resulted in AC invasion failure (arrowhead) and an increase in mig-10b expression in the AC. (C) Quantification of mig-10b expression at the L2/L3 transition, the P6.p 1-, 2-, and 4-cell stages in the animals treated with L4440 control RNAi and egl-43a RNAi (n ≥20 per stage per treatment). (D) The percentage of the ACs that breached the BM at the P6.p 4- (n=85 per genotype) and 8-cell (n=60 per genotype) stages in rrf-3 mutants and rrf-3;mig-10 mutants treated with egl-43a RNAi. Significant differences are indicated (Fisher’s exact test). (E) Within the AC, FOS-1A’s positive transcriptional regulation on egl-43a (arrow), EGL-43A’s negative transcriptional regulation on mig-10b (blunt arrow), and FOS-1A’s positive transcriptional regulation on mig-10b (arrow), form an incoherent (type I) feedforward loop (boxed by red dashed lines). This loop may help maintain a stable, optimal level of mig-10 expression required for effective AC-BM adhesion and AC breaching through BM (blue lines).

To test whether the enhanced mig-10b expression contributes to AC invasion failure caused by loss of egl-43a, we removed mig-10 in egl-43a RNAi-treated rrf-3 mutants that were used as an RNAi sensitive genetic background. Loss of mig-10 suppressed invasion defects in egl-43a RNAi-treated animals (Fig. 3D), indicating that the aberrantly excessive MIG-10B contributes to AC invasion defects resulted from egl-43a RNAi. This incomplete rescue also suggests that the mechanisms independent of enhanced mig-10b expression exist accounting for egl-43a RNAi-mediated invasion failure.

Given that fos-1a positively regulates the expression of both egl-43a and mig-10b during AC invasion [15,17], the revelation of egl-43a’s negative regulation on mig-10b expression unmasks an incoherent (type I) feedforward loop (IFFL) (Fig. 3E), a common motif of gene regulatory networks [26], among fos-1a, egl-43a and mig-10b within the FOS-1A network. As the IFFL is thought to prevent the propagation of potentially harmful fluctuations in gene expression [27], this fos-1a-egl-43a-mig-10b loop may act to ensure mig-10b expression at a stable, optimal level for AC invasion.

3.5 Overexpression of mig-10b disrupts AC invasion and F-actin polarity

To determine whether the expression level of mig-10b is functionally critical to AC invasion, we overexpressed full-length mig-10b cDNA using AC-specific promoter cdh-3 whose activity remains relatively constant throughout AC invasion [19]. Two transgenic lines expressing mig-10b at the low and high levels (2.6-fold difference in fluorescence intensity) were generated. 39% of the ACs (n=14/36) with low mig-10b expression and 63% of the ACs (n=33/52) with high expression failed to invade at the P6.p 4-cell stage, suggesting a potential correlation between the severity of invasion defects and mig-10b expression levels. These two MIG-10B overexpressed lines had a loss in F-actin polarity (Fig. 4B and D), indicating that mig-10b overexpression disrupts F-actin polarity. Supporting this, the egl-43a RNAi-treated animals where mig-10b expression was elevated exhibited a similar polarity loss (Fig. 4C and D). Although it was unclear how overexpressed MIG-10B led to F-actin polarity loss, these results indicate that excessive MIG-10B is detrimental to AC invasion, suggesting the importance of a negative control on mig-10b expression. Such negative regulation may be realized partly through EGL-43A. Further, the negative regulation on mig-10b expression seems to be also required for other biological processes. In the monopolar neurons, such as AVM, mig-10b overexpression causes a multi-protrusive phenotype [12], impairing the unidirectional axon extension. It would be of interest to test whether mig-10b is also transcriptionally regulated in other biological processes in a manner similar to AC invasion.

Fig. 4. Overexpression of mig-10b disrupts AC invasion and F-actin polarity.

(A–C) DIC image (left), corresponding fluorescence (middle), and spectral representation of fluorescence images (right) at the P6.p 4-cell stage. (A) In wild-type animals the AC (arrow, expressing F-actin probe) invaded through BM (arrowhead) and polarized F-actin at the invasive membrane (yellow arrowhead). (B) In the AC overexpressing MIG-10B, the polarized localization of F-actin was loss. (C) In the animals treated with egl-43a RNAi, the AC failed to invade and lost F-actin polarized localization. (D) Quantification of F-actin polarization in the wild-type ACs, the ACs overexpressing mig-10b and the ACs in the animals treated with egl-43a RNAi at the P6.p 4-cell stage (n ≥10 per genotype). Significant differences are indicated (Student’s t-test).

4. Conclusion

Although MIG-10 (Lamellipodin) regulates diverse biological processes [12,13,14], mig-10’s function and regulation in cell invasive behavior are not fully understood. Using AC invasion in C. elegans as an in vivo model, we have found MIG-10B’s synergistic role with UNC-6 in regulating AC-BM adhesion. MIG-10B’s function in cell-matrix attachment is likely associated with MIG-10B promoting F-actin accumulation in the absence of UNC-6. The involvement of MIG-10B in AC-BM attachment extends the function of the FOS-1A transcription regulatory network beyond BM breaching. Further, we unveil egl-43a’s negative regulation on mig-10b expression and suggest the functional importance of the negative expression control on mig-10b during AC invasion. The identification of this negative regulatory link between egl-43a and mig-10b reveals the existence of an IFFL circuit among fos-1a, egl-43a and mig-10b within the FOS-1A transcription network. This IFFL unit may help ensure mig-10b expression at the optimal level required for effective AC invasion.