A proteomic approach identifies timp as a regulator of Collagen IV levels

Given the strong evidence linking Timp to the metabolism of ECM components, we set out to identify ECM proteins that are present in different amounts in control and timp mutant ovaries and to quantitate their relative levels. We performed in duplicate a 4-plex iTRAQ (isobaric Tags for Relative and Absolute Quantitation) analysis of our samples (Fig 1B). Since null timp 1-week old adult ovaries were significantly smaller than sibling controls we decided to use the ovo^D1 dominant mutation, which blocks oogenesis at stage 4–5 of development [22], and thus compare gonads of similar size and containing developing egg chambers of roughly equivalent developmental stages. We used ovo^D1/+;; timp²⁸/TM3 ovaries as control samples and ovo^D1/+;; timp²⁸/Df as experimental tissue (S1A Fig). Utilizing the MASCOT search engine, we identified 644 proteins, which allowed the quantitation of protein products corresponding to 359 genes. We recognized 10 proteins with control/experimental average expression ratios ≤0.50 and 38 with ratios ≥1.50, corresponding to a total of 47 genes (S1 Table). Thus, 13.4% of the identified genes were either under- (2.8%) or over-represented (10.6%) in control versus timp-null ovaries. A bioinformatic search for significant functional pathways and the association of enriched biological annotations to the gene list using PANTHER and GeneCodis tools failed to identify any statistically overrepresented functional pathways or protein networks in the iTRAQ data set. However, the Gene Ontology analysis of the selected candidates indicated several facts. First, three genes with putative proteolytic activity (CG5618, psa and DppIII) were over-represented in timp mutant ovaries. Interestingly, transmission electron-microscopy studies of mutant ovaries showed frequent cellular degeneration not observed in control samples. Escort cells and follicle cells often displayed clear cytoplasms, multi-lamellar bodies and multi-vesicular vacuoles containing cell debris (S4M and S4N Fig). Second and most relevant to the aim of this work, proteins corresponding to nine ECM components were identified by the iTRAQ experiments. Of those, only the two collagen IV α chain homologues present in Drosophila, Cg25C (α1) and Viking (α2) [23, 24] were significantly changed, as they were specifically reduced in experimental tissues (Fig 1C).

In order to confirm that the proteins identified in the iTRAQ study represented changes due to the loss of timp activity, we performed a series of tests. First, we used an LTQ-Orbitrap ion trap mass spectrometer to compare a fraction of the proteome of 1-week old w¹¹¹⁸ (wild-type) ovaries with that of ovo^D1/+;; timp²⁸/TM3 of the same age, as determined by the iTRAQ study. The LTQ-Orbitrap analysis identified 610 proteins with a MASCOT score above 50 and a peptide hit ≥ 2.380 of these (62.3%) were also found in the ovo^D1/+;; timp²⁸/TM3 iTRAQ analysis (S1B–S1D Fig and S1 Table). A Gene Ontology analysis utilizing the GeneCodis tool to search for biological annotations significantly associated to both sets of identified genes rendered similar results, with over 90% of the clustered protein hits in the iTRAQ and LTQ approaches falling in the same Biological Process categories (S1C and S1D Fig). These results corroborated that the iTRAQ experiment identified a representative proteome of the normal tissue. Second, we performed two-dimensional DIfference Gel Electrophoresis (2D-DIGE) to validate proteins differentially expressed either in control ovaries (w¹¹¹⁸;; timp²⁸/TM3) or in experimental ones (w¹¹¹⁸;; timp²⁸/Df). Protein samples isolated from both types of ovaries were labeled with different fluorescent dyes and separated in two-dimensional electrophoresis, which allowed the isolation and identification by mass-spectrometry of three proteins. Two of these were up-regulated in control samples (Chorion protein S16 and Vitelline membrane 26Aa, with standardized logarithm abundance ratios of 3.02 and 3.67, respectively) and the third one was more abundant in timp mutant ovaries (Regucalcin, average ratio of -1.56). In agreement with these findings, the iTRAQ experiment identified another of the chorion proteins in the Drosophila genome (S15) as over-represented in control ovaries, while Regucalcin was found over-expressed in mutant ovaries (S2A–S2C Fig and S2 Table). Third, to rule out the possibility that the genetic combination used in the iTRAQ study (a deletion for the timp gene in trans to a larger deficiency for the locus) could identify as false-positives flanking genes removed by the deficiency or the deletion, we checked individually all the genes removed by both deletion and deficiency (Syn and timp) or by the deficiency alone (CG12814, CG12817, CG12420, CG34107, CG42795, CG3999, CG43143, CG6293, CG12818, CG12592, CG18545, CG31406, Best1, Sirt6, pont, Bruce, sle and jumu). None of the proteins encoded by these genes were identified in the iTRAQ study.

Altogether, the above experiments link the absence of the timp gene and the selective loss of ECM Collagen IV from mutant ovaries, hence suggesting a substantial connection between collagen IV metabolism and Timp activity.

Zymography assays confirm timp as a regulator of Collagen IV proteolysis

To test if timp is involved in the regulation of Collagen IV turnover, most likely by preventing excessive MMP-mediated degradation, we performed an in vivo zymography assay on live ovaries. We incubated 1-week old control and timp mutant ovaries with human Collagen IV-FITC, fixed them, co-stained them with Rhodamine-Phalloidin and measured the intensity of the FITC signal accumulated in the basement membrane of the treated ovaries. FITC fluorescence levels are proportional to Collagen IV-FITC cleavage and the proteolytic activity of the target tissue. Treatment of control and mutant ovaries in parallel rendered significantly different fluorescent intensity values, with the ECM of timp ovaries consistently showing stronger signal levels (Fig 2). Importantly, pre-incubating the samples 30’ with 1mM 1, 10-Phenanthroline—a general MMP inhibitor (see for instance [25]—blocked Collagen IV-FITC degradation (S3 Fig). Thus, assuming that Collagen IV-FITC incorporates at equal levels in both control and experimental BMs, we conclude that Drosophila timp modulates Collagen IV degradation by regulating MMP activity in the ovarian ECM.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. In vivo zymography reveals increased collagenase activity in timp mutant ovaries.

Control and timp mutant ovaries were incubated in culture medium with human Collagen IV-FITC for two hours then fixed and stained to show filamentous actin (F-actin). Green fluorescence corresponds to Col IV-FITC molecules cleaved by zymogen activity in the ovaries. (A) Fluorescence along each ovariole was measured at multiple positions within the following regions: the TF (terminal filament, 12 measurements/ovariole), germarial regions 1-2a (15 measurements/ovariole), and 2b-3 (9 measurements/ovariole), interfollicular stalks (12 measurements/ovariole) and successive egg chamber s (ECh, 12 measurements/ovariole). 5 control and 3 experimental ovarioles were measured. (B) Control ovariole. Note the Col IV-FITC staining decorating the basement membrane and the slightly increased signal in the stalks. (C) timp mutant ovariole. Note that the Col IV-FITC staining, particularly in the anterior tip of the germarium and in the stalks, is stronger than the control. (D) Graph showing the average fluorescence intensity in arbitrary units along the anterior-posterior axis of the ovariole. Images are composites of several focal planes. P-values were obtained using a Student’s t-test. P values <0.05 were considered statistically significant (*:P<0.05, **:P<0.005, ***:P<0.0005). Unless otherwise stated, the genotypes used in this and the remaining figures are: control w¹¹¹⁸;; timp²⁸/TM3 and timp mutant w¹¹¹⁸;; timp²⁸/Df(3R) ED5472. See also S2 and S3 Figs.

https://doi.org/10.1371/journal.pgen.1005763.g002

ECM polarization in older follicles seems unaffected in timp mutants

The germarium is subdivided into 3 regions: Region 1 contains the GSCs and proliferating germ line; Region 2 is populated by 16-cell germline cysts that eventually become enveloped by follicle cells, the progeny of the two follicle stem cells located in this Region; in Region 3 cysts bud off to form developing egg chambers (Fig 1A). To determine if the distribution of major ECM components in anterior ovarian tissues requires timp activity, we studied the distribution of basement membrane components in control germaria by immunostaining. We found strong accumulation of Collagen IV α2 (Vkg), Perlecan, Laminin β-chain and Laminin γ-chain surrounding the terminal filament and GSC niche region. Weaker staining was generally observed along the exterior of Regions 1 and 2, with stronger levels accumulating on the basal side of the nascent follicular epithelia in Region 3 and surrounding the interfollicular stalks. The distribution of the above ECM proteins in timp null mutant females, even as old as 3–4 weeks and exhibiting severe morphological defects, was generally undistinguishable from controls (S4 and S5 Figs). Transmission Electron Microscopy (TEM) analysis on 3–4 week old control and mutant germaria showed an equal accumulation of electron dense material consistent with that observed by confocal microscopy for Collagen IV, Perlecan and Laminins in the GSC niche area and in the lateral regions (S6 Fig).

During mid-oogenesis, the basement membrane of the follicular epithelium serves as a substrate for follicle rotation, a global migration process by which egg chambers execute three complete turns while transitioning from a spherical shape to an elongated egg [14]. During rotation, follicle cells basally secrete matrix components that assemble at the basement membrane, creating a polarized extracellular network perpendicular to the long axis of the developing egg. This “molecular corset” acts to restrict central region growth of the egg chamber while allowing expansion towards the poles [26]. Since timp is required for Collagen IV metabolism, a major component of the basement membrane, we tested whether Collagen IV secreted by the follicle cells could assemble properly at the basement membrane in the absence of timp activity. To this end, we first checked that the polarized accumulation of Collagen IV and Perlecan in rotating egg chambers from 2-week old mutant females was indistinguishable from controls (S4C, S4D, S4J and S4K Fig), suggesting again the absence of gross malformations in ECM organization, at least as judged by confocal microscopy. Second, we performed FRAP (Fluorescent Recovery After Photobleaching) experiments utilizing a GFP protein trap in the viking gene to monitor the deposition of Collagen IV. We measured fluorescence recovery at three different time points, from stages 1–3, when most egg chambers have not yet initiated rotation ([14] and our unpublished observations, but see [27]) and during rotation (stages 5–6 and 7–8). Before stage 5, neither control nor mutant egg chambers showed any sign of recovery in the bleached regions 100 min. after photobleaching. In contrast, stage 5–8 control and timp mutant follicles displayed detectable deposition of Collagen IV:GFP in the photobleached area 100 min. after bleaching (S7 Fig). Since we could not detect statistically significant differences between control and timp mutant samples 100 min. after bleaching, we conclude that the general architecture and organization of the ECM during oogenesis is not grossly affected by the absence of timp function.

Ectopic timp expression results in elongated germaria and egg chamber fusions

The absence of gross alterations in the localization of ECM proteins in mutant germaria could suggest that changes to MMP activity resulting from the lack of timp are highly localized (e.g. affecting only specific cell types) and/or modify ECM properties without affecting its overall distribution. Overexpression of timp in the somatic component of otherwise wild-type ovaries using a UASt-timp insertion [28] combined with either a heat-shock Gal4 or the germarium-specific Gal4 c587 [29] resulted in the formation of compound follicles containing more than one germline cyst (Fig 3A and 3B). This fusion phenotype could be caused by a defect in the encapsulation of the cysts in germarial Region 2 or by a failure in the specification of stalk cells that separate adjacent egg chambers [30]. Significantly, c587-Gal4 expression, which produced the strongest UASt-timp induced phenotypes, is specific to a subset of somatic cells in the germarium [29], implying the need for a fine regulation of extracellular proteolytic activity in anterior ovarian tissues during early cyst formation. Examination of ovaries containing timp-induced egg chamber fusions revealed the presence of cells expressing the stalk cell marker Lamin C towards the periphery of the chambers, in the approximate position where stalk cells would normally be located (Fig 3C and 3D). These observations emphasize the importance of the fine-regulation of timp function in the somatic cells of the germarium for the correct organization of stalk cells and for the encapsulation of new cysts, a process that requires extensive cell migration and ECM remodeling. Importantly, enlarged germaria induced by timp overexpression lack the strong accumulation of Mmp1 in the region where the first interfollicular stalk would form (Fig 3E and 3F; see also Fig 7 for a detailed description of Mmp1 localization), supporting the idea that timp overexpression may affect Mmp1 activity during cyst encapsulation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. timp over-expression impairs cyst encapsulation without affecting cell fate acquisition.

(A, B) Ovarioles from control (no heat-shock) and experimental (daily heat-shock regime for 2 weeks post-eclosion) animals carrying UAS-timp and hs-Gal4. Arrows compare a normal egg chamber containing 16 germ cells with an abnormal, fused egg chamber encasing 32 germ cells. (C, D) Control (c587-Gal4) and experimental (c587-Gal4; UAS-timp) ovarioles grown at 18°C and then switched to 25°C for 9 days. Arrowheads indicate Lamin C-positive stalk cells. (E, F) Control (c587-Gal4) and experimental (c587-Gal4; UAS-timp) ovarioles grown at 18°C and aged 4 weeks at 25°C. Large, empty arrowheads show Mmp-1 accumulation from region 2. The ring of strong Mmp1 staining that coincides with the region contracting to pinch off a new egg chamber in controls is absent in timp-overexpressing germaria (small, empty arrowheads). Occasionally, the TF of experimental germaria show a prominent accumulation of Mmp1 staining, indicating that timp overexpression may affect Mmp1 localization in this region. Images can be composites of several focal planes.

https://doi.org/10.1371/journal.pgen.1005763.g003

Tissue stiffness in the germarium is determined by timp activity

Our data suggest that the dynamics of Collagen IV protein accumulation are affected in the absence of the timp gene. As collagen is an important contributor to tissue stiffness [21], we assessed the mechanical properties of control (n = 7) and mutant (n = 12) ovarioles using Atomic Force Microscopy (AFM)-based indentation measurements. We found a significant reduction in the stiffness along mutant germaria, including the GSC niche and the area where the follicle stem cells reside, as well as in the follicular epithelium of early egg chambers and their interfollicular stalks (Fig 4 and S8 Fig). Our AFM data demonstrate that the absence of timp results in the alteration of physical properties in germarial ECM and early egg chambers. As reported below, the decrease in tissue stiffness may explain the striking morphological defects in tissular organization observed in mutant germaria and their impaired ability to generate new cysts.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. timp mutant ovarioles have altered physical properties.

(A) Subdivision of regions for AFM-based analysis of ovariole tissue stiffness. (B) Comparison of the apparent elastic modulus K at discrete points along 3–4 week old wild-type (n = 7) and timp null mutant ovarioles (n = 12) ex-vivo. Mutant ovarioles exhibited significantly lower levels of tissue stiffness throughout the regions tested. Differences were most severe in the germarium, early egg chambers and their associated interfollicular stalks. Results shown refer to an indentation depth of 0.2 μm. The image is a composite of several focal planes. ** = p values of two-tailed t-tests <0.01. See also S6 Fig.

https://doi.org/10.1371/journal.pgen.1005763.g004

timp is necessary for oogenesis and to maintain ovarian niche tissue stability

Null timp mutant females are sub-viable, semi-sterile and grow smaller ovaries, a phenotype that can be rescued by the leaky expression from a UAS-timp transgene in absence of a Gal4 driver (Fig 5A–5C). Upon closer inspection of 4-week old mutant ovaries, we observed depleted ovarioles containing very few or no maturing follicles (Fig 5D–5F), suggesting that oogenesis is severely affected in the absence of the timp gene and that this depends on the age of the tissue. We thus decided to analyze in greater detail the effects of timp loss-of-function in aging ovaries. First, we focused on the development of interfollicular stalks. While in control ovaries they are formed by a one-cell wide filament of 5–8 cells, timp mutants frequently showed abnormally long interfollicular stalks containing 9 or more cells (Fig 6A and 6B). Lamin C, a marker for differentiated stalk cells, accumulated normally in these additional cells from timp null mutants, suggesting alterations in cell recruitment rather than a non-specific accumulation of cells in mutant stalks. The frequency of ovarioles containing large 9+ cell stalks augmented dramatically with time, increasing from ~15% to ~50% in ovaries dissected from 1 and 7 day old timp females, respectively (S2 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Ovaries from timp mutant females are smaller and lack integrity.

(A) Ovaries dissected from 2-week old control females. (B) Small disorganized ovaries dissected from a 2-week old timp null mutant. (C) Normal ovary morphology restored in ovaries dissected from 2-week old females lacking the endogenous timp gene but carrying a UASt-timp transgene. (D) Germarium from a 2-week old control ovary stained with Rhodamine-Phalloidin to visualize F-actin (red), Vasa to label de germline cells (green) and Hoescht to mark DNA (blue). (E) Germarium from a 2-week old mutant female stained as before. Note the marked decrease in the number of developing cysts within the germarium. (F) Ovary from a 4-week old mutant female stained to visualize the outline of the tissue and DNA. Empty arrowheads point to depleted ovarioles containing very few or no follicles. Arrows: terminal filament. Arrowheads: first interfollicular stalk. Images can be composites of several focal planes.

https://doi.org/10.1371/journal.pgen.1005763.g005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. timp is required to maintain germarium organization and shape.

Distribution of the stalk-cell marker Lamin C in control (A, A’) and timp mutant ovarioles (B, B’). (C) Normal germarium morphology in ovaries dissected from 2-week old control females. FasIII (green) is expressed by pre-follicle cells in the germarium. Phalloidin-labeled actin (red) and DNA (blue) are shown. (D) Abnormal rounded shape of germaria commonly seen in ovaries dissected from 2-week old mutant females. In most cases, despite the altered shape, overall organization of the terminal filament and FasIII expressing cells seem broadly similar to wild-type germaria. (E) Graph quantifying the changing shape of timp mutant germaria 1–7 and 14 days post-eclosion. Differences in length-to-width ratios between control and mutants are statistically significant (p value of two-tailed t-tests <0.001 for 14-day old germaria). (F) Abnormal organization of distinct domains of a timp mutant germarium. In extreme cases, the terminal filament is located to the posterior of region 3, adjacent to egg chambers already pinched off from the germarium. This highly unusual phenotype has not been described previously and is never observed in wild-type germaria. (G, H) The distribution of Escort Cells (labeled in green with the Tj protein) seems normal in aberrantly shaped germaria. The position of GSCs, as determined by their spectrosomes (anti-Hts staining, red) indicates that these cells are still associated to the terminal filament and cap cells, albeit in the example in (H) they are adjacent to follicle cells. Arrows: terminal filament. Arrowheads: first interfollicular stalk. Asterisks: GSC spectrosomes. Images can be composites of several focal planes. See also S2 Table.

https://doi.org/10.1371/journal.pgen.1005763.g006

Second, we analyzed niche organization in detail. Wild-type germaria typically exhibited an elongated shape with a length (anterior-posterior) to width ratio of approximately 2:1 (Fig 6C and 6E and S2 Table). In ovaries dissected from 1–7 day old timp null mutant females, germaria were essentially indistinguishable from controls. In contrast, ovaries from older mutants showed significant alterations to germarium morphology: 59.1% of germaria from 2-week old mutants displayed a much more rounded appearance with a length-width ratio of 1.45:1 (Fig 6D–6F and S2 Table). In such rounded germaria, many gross abnormalities in stem cell niche organization were observed. While control germaria always showed a stereotyped arrangement of cell types, with the terminal filament and cap cells at the tip (Fig 6C), we found mutant germaria where these cells were abnormally positioned. The example in Fig 6F depicts a germarium in which the terminal filament and cap cells are adjacent to an interfollicular stalk and to a differentiated egg chamber. Because the positioning of the GSCs adjacent to the cap cells at the base of the terminal filament is maintained in mutant germaria, the abnormal arrangement observed in a large proportion of aged timp germaria implied that GSCs were positioned in close proximity to follicle cells (Fig 6G and 6H and S2 Table), while the intercalation of Escort Cells in between germline cysts seemed normal. The changes in shape and the alterations in its organization in older mutant adults show that timp plays a role in maintaining germarium structure.

timp is required for proper ovarian niche activity

To analyze the outcome of niche activity, we determined whether loss of timp affected the ability of mutant ovaries to produce new germline cysts. We counted the number of 2-, 4-, 8- and 16-cell cysts present in 10-day and 3-week old germaria. Compared to controls, timp mutants showed a clear diminution in the number of cysts populating 3-week old germaria (controls 9.5±2.2 cysts/germarium, n = 19; mutants 5.8±3.5 cysts/germarium, n = 29. S3 Table). This difference is not a consequence of a drop in the number of GSCs present in the niche, as control and timp hemizygous germaria did not show statistically significant differences in GSC numbers either 10 days or three weeks after eclosion (S3 Table). This result indicates that Collagen IV degradation in mutant tissues does not affect visibly niche signaling. Given the progressive loss of developing cysts in mutant germaria even in the presence of a normal pool of GSCs, we next performed a lineage experiment in which we compared the ability of control versus mutant germaria to produce new cysts. To this end, we generated flip-out clones to induce the expression of a reporter gene (mCD8::GFP) in random germline cysts from 1-week old ovaries. Females of this age were chosen because the vast majority of mutant ovaries this old do not present gross morphological alterations that might cause pleiotropic effects on cyst production (see above). We scored the number of germaria containing GFP-expressing cysts 3 days after clone induction to make sure they were generated from labeled GSCs or cystoblasts. We found marked germline cysts in 18.5% of control ovarioles (n = 216) and in 10.4% of mutant ones (n = 202). Normalized by the total number of GSCs in each of the groups, we found a frequency of 6.6 marked ovarioles per 100 GSCs in control ovaries, while mutant ovaries contained on average 4 marked ovarioles per 100 GSCs (S4 Table). Altogether, the above results demonstrate that 1-week old timp mutant germaria are only 60.7% efficient in germline cyst production compared to controls of the same age and they strongly support the idea of a direct correlation between timp activity and tissue homeostasis.

timp regulates Mmp1 and Mmp2 distribution

Because the canonical role for Drosophila timp is to regulate MMP activity in a variety of tissues, we sought to determine if MMP distribution was affected in mutant ovaries. Mmp1 localizes to the terminal filament and in the cap cells area in control ovaries. From region 2 onwards it accumulates evenly along the basal side of the follicle cells. Stronger signal can be detected in the interfollicular stalks of stage 2 and older egg chambers. In striking contrast, null timp mutant germaria and interfollicular stalk cells show a diffused distribution of Mmp1 protein (Fig 7A and 7B). An Mmp2::GFP fusion protein driven by its endogenous promoter has been reported to accumulate most strongly in the anterior tip of the germarium, where it mediates the distribution of the niche signal Wingless [10]. Upon examination of the localization of Mmp2::GFP in mutant germaria, we found that timp mutants do not show a conspicuous signal in the niche region, indicating that, like in the case of Mmp1, timp activity affects Mmp2 localization (Fig 7C and 7D). This view is supported by the fact that timp mRNA is expressed strongly in the area were GSCs reside and in region 2 (Fig 7E and 7F). Interestingly, the expression domains of Mmp1 and Mmp2 in the germarium and of Mmp1 in the interfollicular stalks correlate with the timp mutant phenotypes described above and suggest a possible role for timp in regulating MMP activity in the germarium. Because the germarium and interfollicular stalk phenotypes are rescued by the ectopic expression of timp (see Materials and Methods and S2 Table), our observations suggest that endogenous timp might antagonize Mmp activity that to ensure germarium organization and proper interfollicular stalk formation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Mmp1 and Mmp2 are abnormally distributed in timp mutant ovaries.

(A, A’) Pattern of Mmp1 accumulation in control ovaries. (B, B’) timp null mutant ovaries (timp²⁸/Df ED5472) displayed a diffused Mmp1 staining, particularly in the stalks. (C, C’) Mmp2::EGFP is localized to the anterior half of the germarium in control ovaries. (D, D’) Mmp2::EGFP localization at the anterior tip of timp mutant germaria was not detected. Arrows: terminal filament. Arrowheads: interfollicular stalk. (E, E’) Double in situ hybridization and antibody staining to visualize the pattern of expression of timp mRNA in a wild-type germarium (E and red signal in E’) and the spectrosome marker Hts (green in E’). DNA was counterstained in blue. timp was strongly expressed at the tip of region 1, were GSCs and cystoblasts reside, and in region 2. Images can be composites of several focal planes.

https://doi.org/10.1371/journal.pgen.1005763.g007

timp regulates Collagen IV turnover in the Drosophila ovary

Most evidence from Drosophila has reinforced the canonical view of MMPs and Timp as opposing regulators of ECM turnover, especially in tissue remodeling [28, 32]. For example, in a number of different Drosophila tissues ectopic timp expression phenocopies the loss of MMPs [33–35] and the phenotypes of the sub-viable timp mutant adults used in this work, such as wing blisters [12], are strongly suggestive of perturbed ECM regulation. However, up to now little or no direct evidence has been presented for the roles of endogenous timp in Drosophila or how its activity impinges on the regulation of ECM turnover. In this work we present several lines of evidence that support a canonical role for Timp in regulating ECM turnover in the Drosophila ovary. The reduced levels of the two Collagen IV subunits in timp mutant ovaries are consistent with an alteration in the turnover of this conserved basement membrane component [36]. Collagen IV is composed of long helical trimers that form a network via end domain and lateral interactions [37]. Significantly, because the increased proteolytic cleavage of human Collagen IV-FITC by cultured timp mutant ovaries could be blocked using a general MMP inhibitor, our observations strongly suggest that the loss of timp alters Collagen IV turnover via changes to MMP proteolytic activity. Drosophila embryos with reduced MMP activity showed lower levels of Collagen IV accumulation at the leading edge of epidermis wound sites [38]. Because Mmp1 and 2 act redundantly to promote wound healing, which could be blocked by overexpressing timp, MMP activity is required for Collagen IV deposition at least in the embryo. Our results indicate that during oogenesis endogenous timp activity is required for normal ECM organization, most likely by maintaining an appropriate balance of extracellular protease activity, a suggestion in agreement with the fact that timp overexpression blocks MMP2-dependent posterior follicle cell trimming during ovulation [35]. Further investigation will be required to confirm if one or both of the Drosophila MMPs are required for Collagen IV remodeling in oogenesis. However, it is also possible that timp controls ECM composition in other ways, as MMP-independent functions have been reported for timp in other organisms (reviewed in [39]).

timp determines the mechanical properties of the ovarian niche

Given the altered Collagen IV levels in timp mutants and the morphological changes to ovariole organization, we expected to see alterations to ECM composition and gross basement membrane lesions. Surprisingly, we found that the distribution of conserved basement membrane components was not consistently affected when examined by confocal microscopy. Nor did we observe significant alterations to the strikingly electron dense extracellular matrix accumulations seen in the apical stem cell niche regions of the germarium. Atomic Force Microscopy has proved useful for studying the regulation of ECM molecules by extracellular proteases in other models (reviewed in [40]). Our examination of the mechanical properties of control and timp mutant ovaries using this method demonstrated that germaria and egg chambers were significantly softer than control tissues. While ovarioles are composite materials with different structures contributing to the measured apparent elastic modulus (K), structures closer to the surface (such as the basement membrane) will contribute more than internal components. In fact, structures further away from the surface than ~1/10 of the indentation depth will not add to K [41, 42]. Given the ~100nm thickness of most ovarian basement membranes, the dramatic reduction in tissue stiffness of timp mutants in the 200nm indentation measurements is almost certainly a consequence of changes to the basement membrane itself. Under normal circumstances and considering the role of Collagen in determining a tissue’s elastic stiffness [21], the consistently Collagen IV-rich ovarian basement membranes are likely to form a stiff layer. This conserved ECM component has been shown to promote the basement membrane-mediated constriction of Drosophila tissues and organs [43]. Overall our data strongly suggest that in the absence of timp the ECM surrounding ovarian tissues, and particularly the germarium, has altered mechanical properties conferring lower stiffness despite a largely normal distribution of major ECM components.

The nature of ECM changes in timp mutants remains to be determined. It could reflect alterations to the molecular organization of ECM components such as Collagen IV and Laminin that are not visible to confocal or electron microscopy. Although we have tested the distribution of many different ECM components in Drosophila, it was not possible to examine the distribution of all the 20+ known or putative ECM components encoded by the Drosophila genome [44]. Developing suitable tools to determine the distribution of these additional ECM components will be needed to fully evaluate the role of timp in determining ECM composition. Of particular interest will be the remaining uncharacterized putative ECM components, such as Glutactin [45], that were detected by the iTRAQ screen as being expressed in the ovary.

Is MMP activity in the ovary controlled by redundant mechanisms?

The relatively normal distribution of ECM components in timp mutant ovaries is related to another key question: how do adult flies develop and survive for weeks in the absence of timp? What stops secreted and membrane-bound MMP proteins from wreaking havoc in the absence of Timp? Mouse knockouts of individual timp genes are viable [46–49] and often fertile, exhibiting relatively minor defects in specific tissues frequently alleviated by decreasing MMP function in the affected tissues (reviewed by [50]. There are some interesting parallels between the effects of removing timp from mice and Drosophila. For example, mice null for timp-3, the closest orthologue to Drosophila timp [11], are viable but show decreasing lung function (and increasing collagen proteolysis) with age that eventually results in death after around 13 months [48].

MMP activity is believed to be tightly regulated at multiple levels including transcription, secretion and zymogenic-activation [51]. Our observation that MMP accumulation in the basement membrane is reduced in timp mutants suggests that the expression of Mmp1 and Mmp2 might be regulated to compensate for the absence of timp. In other Drosophila tissues, it has been shown that the JNK pathway controls Mmp1 expression [34, 52]. We hypothesize the existence of feedback mechanisms in the ovary that detect the level of extracellular proteolytic activity and adjust MMP expression accordingly, a mechanism that may alleviate the absence of timp function in younger mutant females but that is insufficient to compensate the prolonged loss of timp in aged ovaries. The eventual defects seen in timp mutants as they age might result from less effective tissue homeostasis caused by the long-term absence of Timp protein or by reduced MMP expression.

The role of timp in the ovarian stem cell niche: a close link with MMP expression

The germarium houses germline (GSC) and follicle stem cell (FSC) populations that give rise to the germline and somatic components, respectively, of developing egg chambers. Cap cells, escort cells and terminal filament cells are responsible for producing niche signals that permit the maintenance of GSCs in their undifferentiated state [19]. Here, we have shown that the loss of timp can severely alter the shape and organization of the different domains of the germarium as mutant flies age and enlarge interfollicular stalks between egg chambers. Over-expression of timp in the germarium produced the opposite effects with elongated compound germaria and egg chamber fusions. In spite of the often-dramatic changes to germaria morphology seen in timp mutants, the basic association of GSC, cap cells and the terminal filament was always intact. This is consistent with our observation that GSC number is unaffected in timp mutants and suggests that E-cadherin-mediated adhesion between GSCs and cap cells [53] maybe sufficient to maintain local GSC niche integrity even if ECM properties are compromised. Thus timp appears to be important for the long-term maintenance of germarial structure and its stereotyped arrangement of different subdomains.

FSC maintenance and proliferation in the niche depends on binding Laminin that they themselves secrete [54]. In addition to providing structural support within stem cell niches, the ECM is also believed to play a key role in stem cell niche signaling [2, 21]. Recently, it has been shown that the secreted glypican Division abnormally delayed (Dally)-like (Dlp) promotes the long-range action of the Wingless (Wg) ligand in the germarium. Mmp2, which can cleave Dlp in cell culture, opposes Dlp activity in the ovary, thus limiting the range of Wg signaling. Interestingly, the phenotypes associated with loss- and gain-of-function conditions for Mmp2 are similar to those we identified for timp, with long stalks seen in conditional Mmp2 mutants and egg chamber fusions when overexpressed [10]. These results are in accordance with our finding that loss of timp results in decreased Mmp2 accumulation in the anterior tip of the germarium and suggest that timp activity may regulate FSC proliferation via Mmp2. In support of this idea, ectopic timp overexpression suppressed both the long stalks of Mmp2 mutants and egg chamber fusions of Mmp2 overexpression [10].

We envisage the basement membrane as a physical corset responsible for the maintenance of the proper shape of the ovarian niche, in spite of the tensions generated by the cell divisions and cellular movements that occur during new egg chamber assembly. This corset could be analogous to the Collagen IV—dependent structure that controls follicle shape during of the later stages of oogenesis [14]. The circumferential migration of FSC daughter cells [55] may play a role in secreting and/or organizing such a circular structure. Considering the striking defects observed in the organization of the mutant ovarioles, in which germaria displayed abnormal shapes and an aberrant arrangement of cell types, we propose that the loss of Timp regulation of the ECM causes a softening of the basement membrane, which is now unable to act as the corset upon which the tissue is modeled. As a consequence and as indicated by the progressive loss of developing cysts in aging mutant germaria, the new arrangement of the GSC and FSC niches provokes a significant impairment of ovarian homeostasis. Because a large number of stem cell niches of both vertebrates and invertebrates have been found to be contained within, or to interact with, specialized ECMs, the findings reported here provide a link between ECM metabolism, niche organization and tissue homeostasis that may be of general importance for the biology of stem cells.

Fly stocks

We generated a timp null condition by combining a ∼15 kb deletion of that removes both timp and synapsin (timp²⁸, a gift from A. Page-McCaw) with the larger Df(3R)ED5472 (Bloomington Drosophila Stock Centre). Females bearing a deletion only affecting synapsin (syn²⁷) [12] over Df(3R)ED5472 display none of the phenotypes associated with the removal of timp. The presence of a UASt-timp and either Heat-Shock Gal4 or the c587-Gal4 transgenes in timp²⁸/Df(3R)ED5472 mutant females grown at 25°C was sufficient to significantly restore normal ovary morphology. Viking: GFP is a protein trap in the endogenous Collagen IV α2-chain (line G00205; FlyTrap; http://flytrap.med.yale.edu/). UASt-timp, a gift from A. Page-McCaw, is a pUASt insertion carrying the complete timp coding sequence [28]. c587-Gal4 is a germarium-specific GAL4 line [29]. Mmp2-EGFP is an engineered BAC construct (P[acman]-Mmp2-EGFP-GPI; [10] that expresses EGFP-tagged Mmp2 under endogenous regulatory sequences.

To perform the flip-out experiments, we obtained w, hsFLP12/w; 10xUASt-mCD8::GFP/+; timp²⁸ Act>y⁺>Gal4/TM6B and w, hsFLP12/w; 10xUASt-mCD8::GFP/+; timp²⁸ Act>y⁺>Gal4/Df(3R)ED5472 genotypes. Control and experimental flies were heat-shocked at 37°C for 30 minutes and the ovaries dissected and processed for immunostaining 3 days later.

Preparation of fixed ovaries for observation. Antibodies used

Antibody, DNA and rhodamine-phalloidin stainings were performed according to standard procedures. A list of the antibodies used is available in the Supporting S1 Text.

Preparation of live ovaries for observation

Individual ovarioles were dissected in Schneider medium supplemented with streptomycin and insulin as described previously [56].

Fluorescence Recovery After Photobleaching (FRAP)

Regions of interest were bleached at 100% 488 nm laser power for three 2-second scans (400 Hz, one line/frame average). Images were then collected at 20-minute intervals for 2 hours.

Quantification of fluorescence

Image series were captured using identical confocal settings for control and experimental ovarioles. Color depth was set to 12-bit and configured to minimize saturated pixels. Depth (z) thresholds were set well above and below each ovariole to guarantee the complete tissue was captured. Sections were taken at 630 nm intervals (optimal). FITC was captured using the default Leica FITC configuration and low laser intensity. After recording position and z-depth ranges, image series were captured automatically. All image stacks were pre-processed using the standard background subtraction function of ImageJ (default settings; 50 pixel radius). Measurements were taken using the Imaris Measurement Points Tool along the length of each ovariole (4 in the terminal filament, 5 in regions 1-2a of the germarium, 3 in regions 2b-3, 4 in each of the stalks and 4 measurements in each of the egg chambers). Each ovariole contained at least 3 stalks and 4 egg chambers. Ovarioles were measured along three different lines running from the germarium to posterior follicles. Thus, TFs were quantified 12 times/ovariole, the niche region 6 times/ovariole, etc.

Transmission Electron Microscopy (TEM)

TEM samples were prepared following standard procedures. Sections were examined with a Zeiss EM902 electron microscope at 80Kv, and photographed at 50.000x magnification.

Collection of Drosophila ovaries for proteome analysis

0–3 day-old females were yeasted for 3 days prior to dissection. Ovaries were dissected in PBS and immediately frozen in liquid nitrogen. For the iTRAQ experiment, we collected four biological replicas. For the LTQ-Orbitrap, we analyzed two biological replicas. For the 2D DIGE analysis, we collected four biological replicas.

iTRAQ quantitation

Protein quantitation was carried using four independently-collected samples from control and experimental ovaries. A MALDI TOF/TOF 4800 (AB SCIEX, Foster City, CA, USA) mass spectrometer was used for acquisition and data processing. An overview of the experimental design is shown in Fig 1B. The final quantitation was performed on identified proteins associated at least to three quantitated peptides.

LC-MS/MS analysis of wild-type ovaries on a LTQ-Orbitrap

Prepared protein samples were digested with 1:20 sequencing grade trypsin (Roche Molecular Biochemicals). Peptides were analyzed using a linear trap Orbitrap Velos (LTQ Orbitrap Velos) hybrid mass spectrometer. The identified fragments were searched against the Drosophila melanogaster database of UniProtKB/Swiss-Prot using Mascot (version 2.3.0).

DIGE analysis

A total of 50 μg of proteins from each condition labeled with 400 pmol of Cy3 or Cy5 dyes were used to perform Isoelectro focusing. The second dimension was performed on 12% SDS-PAGE gels. Fluorescent gel images were scanned and changed protein spots with p-values ≤0.05 were picked for finger-printing identification. MALDI samples were automatically analyzed in an Ultraflex MALDI-TOF/TOF mass spectrometer.

GO analysis of the differentially expressed proteins

To perform ontological and functional studies, a list of candidate genes coding for differentially-expressed proteins was evaluated using PANTHER (http://www.pantherdb.org/) and GeneCodis (version 3; http://genecodis.cnb.csic.es/) tools.

Zymography assays

0–3 day-old females were yeasted for 3 days prior to manipulation. To assay for the collagen IV degradation in live tissue, we dissected control and experimental ovaries in Schneider’s medium supplemented with Fetal Bovine Serum (15% vol/vol) containing streptomycin/penicillin [56] and incubated them in a culturing cocktail containing Collagen IV-FITC (Collagen, type IV from human placenta, fluorescein conjugate; Invitrogen). Upon fixation in 4% paraformaldehyde, the fluorescence of control and experimental samples were captured using identical confocal settings and measured using IMARIS software.

Atomic Force Microscopy analysis

137 measurements from 7 ex-vivo wild-type ovarioles and 179 from 12 timp nulls were collected. Monodisperse polystyrene beads were glued to silicon cantilevers with a nominal spring constant of 0.1 N/m. Force-distance-curves were taken at an approach speed of 10μm/s and a maximum force F = 6nN. Force—distance curves were analyzed for different indentation depths δ (0.2μm, 0.5μm and 1μm) using a Matlab-based custom algorithm [57].