Aggrecan Directs Extracellular Matrix-Mediated Neuronal Plasticity

Animal housing and ethics.

All animal work was performed in accordance with national law and regulation, in Cambridge in accordance with the UK Animals Scientific Procedures Act (1986), and in Oslo by the approval of the Norwegian Animal Research Committee. Animals used in this study were housed in standard housing conditions with a 12 h light/dark cycle. Animals were housed in groups of two to five per cage, with the exception of one individual animal. Animals were fed and watered ad libitum.

Generation of Acan-loxP and breeding strategy.

Two JM8A3.N1 embryonic stem (ES) cell lines of agouti C57BL/6N origin (Pettitt et al., 2009); mice #HFP0602_5_G11 and #HFP0602_5_F12, carrying the targeted Acan gene, were purchased from the European Mouse Mutant Cell Repository (EuMMCR). The two ES cell lines are clones containing the same transgenic cassette surrounding exon 4 of the mouse Acan gene (Fig. 1a) a general “knock-out first” design used by the EUCOMM project, termed transgenic allele tm1a (Skarnes et al., 2011). Genomic DNA from the ES cell lines was analyzed by Southern blotting to confirm correct genetic targeting. ES cells were plated and prepared according to the supplier guidelines before microinjection into 50 blastocysts of albino C57BL/6J tyr^−/− origin and subsequent transfer to three pseudopregnant females per line. The chimeric offspring, hybrid F₁ generation, was selected by the appearance of coat color spots originating from transgenic cells. The F₁ animals were then crossed with albino C57BL/6J tyr^−/− mice, and the transmission of genotype into the G₁ generation was detected by coat color and verified by PCR genotype analysis using GT3 and wild-type (WT) primers (Table 1), confirming the heterozygote tmc1a transgenic animal. The tm1a allele is not viable in a homozygous state, as the transgenic cassette interferes with Acan expression. The mouse was further crossed with the ROSA26::FLPe strain expressing FLP-FRT recombinase removing the majority of the transgenic cassette leading to off spring carrying the Cre-lox conditional knock-out allele tm1c (Fig. 1a), designated as the conditional knockout of Acan; B6(Cg)-ACAN^{tm1c(EUCOMM)Hmgu>/Jwfa}, hereafter referred to as Acan-loxP. PCR genotype analysis using primer pairs FLP, tmc1, and GT5, confirmed presence of FLP recombinase and presence of tm1c. The tm1c allele is viable as a homozygote. Conditional knockout of Acan is achieved by removing exon 4 of the Acan gene by Cre-lox recombination, resulting in the tm1d allele (Fig. 1a). Loss of exon 4 induces a shift in the reading frame, and thereby prevents protein translation.

Brain-wide and developmental Cre-lox recombination by Nestin-CRE.

To achieve brain-wide knockout of Acan, Acan-loxP^+/+ was crossed with Nestin-CRE (CreNes), where Cre recombinase is expressed under the control of the promoter and the nervous system-specific enhancer present in the second intron of the rat Nes gene (Tronche et al., 1999). The offspring of Acan-loxP^+/−CreNes mice were then backcrossed for at least one generation. The offspring of this cross, WT littermates, Acan-loxP^+/+, Acan-loxP^+/− Cre, and Acan-loxP^+/+ Cre, were used for histological analysis and behavioral testing. Again, PCR genotype analysis was conducted using WT, GT3, GT3WT, and CRE primer pairs (Table 1). All of the resulting offspring had a normal phenotype and body size (data not shown). Juvenile animals explored and developed normally and were able to rear and feed themselves (experimenter observations). All genotypes were established using PCR on genomic DNA extracted from an ear punch.

Local and acute Cre-lox recombination by virus and retrograde tracer injection.

AAV9.hSyn.HI.eGFP-Cre.WPRE.SV40 (catalog #AV-9-PV1848) was purchased from the Penn Vector Core (University of Pennsylvania, Philadelphia, PA), and stored at −80°C. Before surgery, aliquots were thawed on ice and diluted in filtered 1× PBS to a concentration of 10¹³ GC/ml. Adult mice (3–4 months old) of either Acan-loxP^+/+ or Acan-loxP^+/− genotype were anesthetized with isoflurane mixed with air (5% induction, 1.0–1.8% maintenance) using a Somnosuit vaporizer (Kent Scientific) and a custom-made mask, and were placed on a controllable heating pad in a stereotaxic frame. They were injected subcutaneously with buprenorphine (0.04 mg/kg) and carprofen (5 mg/kg), and bupivacaine (13 mg/kg) locally. The scalp was shaved and cleaned with 70% ethanol and chlorhexidine, and a small incision was made in the skin. A craniotomy (diameter, 1–1.5 mm) was performed above the binocular zone of primary visual cortex (V1) using a handheld dental drill. The virus solution was loaded into a glass pipette (opening diameter, ∼15 μm) and mounted in a Nanoject 3 Microinjector (Drummond Scientific). A total of 500 nl was injected in a stepwise manner over 10 min, and the pipette was left in the tissue for 10 min before retraction [injection coordinates relative to lambda were: anteroposterior (AP), 0.0 mm; mediolateral (ML), 3.0 mm; dorsoventral (DV), 0.5 mm]. The wound was rinsed repeatedly with sterile 0.9% NaCl and closed by suture. The animals were observed for at least 1 h before they were placed back in the colony. Animals were monitored closely for a week following surgery, and were given subcutaneous injections of buprenorphine and carprofen for the first 3 d after surgery. For retrograde tracing between V1 and the dorsal part of the lateral geniculate nucleus, cholera toxin subunit B conjugated to Alexa Fluor 488/594 was purchased from Life Technologies and a 1% solution was made by reconstitution in 1× PBS. Surgical procedures were as described above. The tracer injections were intentionally targeted to a more medial part of V1 to prevent leakage into the neighboring V2L region. A total of 200 nl was injected for each of the tracers at 0.0 mm AP, 2.7.0 ± 0.1 mm ML, and 0.5 mm DV, and the injections of each conjugate were separated by 0.2 mm. The wound was cleaned with sterile 0.9% NaCl and closed by suture. Medication was administered as described above.

Histology.

At the end of the experiment or an appropriate time point, animals were deeply anesthetized by an intraperitoneal injection of sodium pentobarbital (50 mg/kg) and perfused with either 0.9% NaCl [ocular dominance (OD) experiment] or Na₂HPO₄.2H₂O (Acan-loxP CreNes) buffer, followed by 4% paraformaldehyde (PFA). Brains were dissected out and postfixed in PFA solution overnight and cryoprotected in 30% sucrose in 1× PBS for 24 h. Coronal sections of 30 or 40 μm were collected from the cryostat or freezing microtome and transferred to 1× PBS. Sections were rinsed in 1× PBS and blocked for 1 h in 2–10% animal serum (species dependent on secondary antibody host) before primary antibody incubation overnight in blocking solution (Table 2). The following day, sections were rinsed and incubated with secondary antibodies for 2 h. Sections were rinsed and mounted on Superfrost slides with FluorSave Reagent (Merck Millipore).

Following OD testing of AAV9.Cre-injected Acan-loxP^+/+ mice, wide-field fluorescence images were acquired with an AxioPlan 2 Microscope (Zeiss) through a 10× objective, and high-resolution overview images stitched using the Mosaix extension in the AxioVision Software. Confocal images were acquired using a FluoView FV1000, software version 1.7 (Olympus), through a 20× or 60× PlanApo objective. Image analysis was performed using ImageJ (NIH) and Photoshop CS4/6 (Adobe).

For histological quantification of 3- to 4-month-old Acan-loxP^{+/+ ×} CreNes crosses, images were captured using a Leica SPE confocal microscope using either 10× or 63× objectives with a 1024 × 1024 image resolution (n = 3/group). For PNN quantification, at least five images per section (three sections per brain, ∼300 μm apart) were taken on a single confocal plane in layers 4 and 5 of the mouse barrel cortex (Fig. 2). Images contained at least one Wisteria floribunda agglutinin (WFA)-labeled PNN (if detected). For Acan-loxP^+/+ CreNes, where no PNNs were detected, representative images were captured randomly in the same region. For parvalbumin (PV) cell analysis by immunohistochemistry at least five z-stack images (10 stacks, ∼3 μm apart) were taken per section with at least five sections analyzed per animal (∼300 μm apart). Images were then analyzed with an automated custom script using the EBImage package from R statistical software (Pau et al., 2010). To reduce variance in antibody binding, all groups used for quantification were histologically stained in batches using the same conditions and antibody solutions.

Protein and RNA analysis.

For aggrecan protein and gene expression analysis, snap-frozen Acan-loxP^+/+ and CreNes Acan-loxP^+/+ brains (n = 3/group) were homogenized over ice, and both protein and RNA were extracted using an AllPrep DNA/RNA/Protein Mini Kit (Qiagen). Before aggrecan protein quantification, 20 μg of brain homogenate from each sample was deglycosylated with chondroitinase ABC (ChABC) in PBS with acetic acid (50 U/ml, pH 7.8) for 24 h at 37°C to remove GAG side chains from the aggrecan core protein. Following deglycosylation, samples were subjected to the SDS-PAGE method using a 4–12% Bis-Tris Mini Gel (NuPAGE, Invitrogen) and Western blotting protein transfer to a PVDF membrane. Membranes were left to air dry before being washed three times with 2% Triton X-100 in Tris-buffered saline–Tween 20 (TBS-T). After the final wash, membranes were incubated in a blocking buffer (10% skimmed milk made in TBS-T) for 1 h. After blocking, membranes were then incubated in primary antibodies overnight at 4°C with 5% skimmed milk in TBS-T. The following morning, membranes were washed in triplicate in TBS-T and incubated in secondary antibody for 1 h, after which they were washed again. Proteins were then visualized using the electrochemiluminescence method. Protein quantification was achieved using ImageJ comparing aggrecan signal intensity relative to β-actin. For aggrecan, all bands >250 kDa were included for analysis similar to previous studies (Miyata and Kitagawa, 2016), while β-actin produced a clear band at 37 kDa.

For gene expression analysis, mRNA was converted to single-stranded cDNA using a High-Capacity cDNA Reverse Transcription Kit with RNase Inhibitor (Applied Biosystems). Samples were then analyzed using quantitative PCR with inventoried TaqMan probes (Applied Biosystems) for Acan and Actb. A final concentration of 14 ng of cDNA per reaction was used. All samples were run in triplicate and averaged before analysis. Measurements for the expression of Acan were then presented relative to Actb for graphing and statistical analysis.

Quantification and statistics of the histology signal.

Before image processing, a 30% threshold was applied to the image to eliminate nonspecific binding. Cells whose soma were in the z-stack image were extracted, combined intensity values for all pixels in a cell were averaged, and cropped images of PV⁺ cells were saved for manual verification. The Arbitrary Unit (AU) scale was derived as before based on mean PV values extracted from the WT group.

Assessing OD plasticity in visual cortex using optical imaging of intrinsic signal.

Visual stimulus was generated in Matlab using the Psychophysics toolbox, and presented on a 27.5 × 34 cm computer monitor (Dell Computers; 60 Hz refresh rate) placed 25 cm from the mouse. The stimulus consisted of a thin drifting bar (2° wide) moving up or down with temporal frequency of 0.17 Hz and spatial frequency of 0.05 cycles/°. The stimulus was repeated 20 times, with each presentation interleaved with a gray blank screen. The duration of each recording was 120 s. Stimulus was restricted to the binocular field of vision (20°), with an offset of 5° toward the respective open eye.

Eight to 10 weeks after virus injection, the animals were prepared for imaging. The experimenter performing imaging experiments and analysis was blind to treatment group and genotype. A custom steel plate for head fixation was attached to the skull using cyanoacrylate and dental acrylic, leaving the area above the visual cortex open. This was covered by a thin layer of nitrocellulose (New-Skin) to prevent desiccation and reactive cell growth.

After head plate implantation, the first imaging of intrinsic signals was performed to measure baseline responses through each eye. Repeated optical imaging of intrinsic signals was performed as described previously (Kaneko et al., 2008). The mouse was anesthetized with isoflurane (2% for induction and 0.5% during recording) supplemented with an intramuscular injection of chlorprothixene (2 mg/kg body weight). The mice were head fixed and placed on a heating pad in front of the computer monitor, and images were recorded transcranially by a Dalsa M30P Camera (Teledyne) using two adjacently mounted Nikon 50 mm f/1.2 lenses, and custom Matlab software. Light was provided by LED lights mounted in a ring surrounding the camera lens. Two 525 nm LEDs were used to focus on the surface blood vessels before focus was shifted 600 μm below the surface. The light source was switched to three 615 nm LED lights (XPEBRO-L1-R250–00B02, Cree), and emission light was filtered with a red interference filter (610 ± 10 nm). Images were acquired with a frame rate of 30 frames per second. The phase and amplitude of cortical responses at the stimulus frequency were extracted by Fourier analysis, as described previously (Kalatsky and Stryker, 2003). The responses to the stimulation of each of the eyes were recorded for four to six trials, alternating between which eye was covered. The median of the peak response amplitude (R) of all trials for each eye were used to calculate an OD index (ODI) by ODI = (R^contra − R^ipsi/R^contra + R^ipsi).

Immediately after baseline imaging, the eyelids to the contralateral eye was sutured shut. Lidocaine was applied to the sutured area, and the animal returned to its home cage. After 4 d, the mice were again anesthetized, the closed eyelid was reopened, and the imaging procedure was repeated as described above.

Assessing memory by spontaneous object recognition behavioral task.

To gauge the memory ability of Acan-loxP^+/+ Cre mice, an SOR behavioral task was used. During testing, male mice (n = >7 per group, ∼3 months old) were placed individually in a Y-shaped maze (30 cm high, all arms 16 cm in length and 8 cm wide) as previously described (Romberg et al., 2013). Briefly, animals were habituated to the testing arena, where animals were allowed to explore freely for 5 min each day, for 2 consecutive days before testing. Each testing session occurred during the dark phase and consisted of a sample phase and a choice phase, using the same objects as in the study by Romberg et al. (2013). During the sample phase, a mouse was placed in the start arm of the maze with sample objects (two identical objects) placed in the two remaining arms and left to explore the objects freely for 5 min. After a 3 or 24 h delay, animals were returned to the maze for the choice phase. During the choice phase, two new objects were placed in the sample arms, one that is the same as the sample objects (familiar object) and one that was new to the mouse (novel object). For each delay time, mice underwent at least one testing session. All testing sessions were separated by a minimum of 48 h to allow recuperation. The presentation of the object pairs and their appearance in either the left or right sample arm were counterbalanced so that a particular object in a set or a particular side might be novel for one animal and not for another to avoid unforeseen bias. The maze and all objects were wiped down with 70% EtOH in between each testing session to avoid any olfactory bias that could have been introduced during the previous testing session. Time spent exploring each object during the sample and choice phase was assessed and scored at a later date using the video recordings. A discrimination score was calculated for the choice phase by subtracting the amount of time exploring the familiar object from the amount of time exploring the novel object and then dividing by the total exploration time, as follows: A positive score indicates recognition of the novel object, while a score of ≤0 indicates the mouse was unable to recognize the novel object. An average was taken for the total sample times per object and the discrimination scores for the choice per time delay.

Experimental design and statistical analysis.

Statistical analysis was performed using R (R Foundation) or GraphPad Prism version 7 (GraphPad Software). Datasets were tested for normality using the Shapiro–Wilk test or Fisher's F test. Comparisons between two groups were tested by Welch's t test or paired t test. Multiple group analysis was achieved using a type 1 or type 3 ANOVA with the appropriate follow-up test, either post hoc Tukey's HSD or comparing two groups by Welch's t test.

Biochemical quantification of aggrecan protein and the Acan transcript were conducted as described in the Protein and RNA analysis section above. Separate brain homogenates from three male animals, 3–4 months of age from litters originating from the same breeding pair, belonging to each of the two experimental groups, Acan-loxP^(+/+) and Acan-loxP^(+/+)Cre, were analyzed. For protein quantification, ImageJ was used to quantify the aggrecan signal, which was normalized to β-actin signal, with each lane corresponding to an individual animal (Fig. 1d). For mRNA quantification, triplicates of 14 ng of cDNA derived from each of the individual animals were run with TaqMan probes against Acan and Actb, and the relative signal was quantified by the accompanying software (Fig. 1e). The two experimental groups were compared and tested in a Welch's t test (Table 3).

Quantification of PNN intensity by WFA and aggrecan immunohistochemistry were conducted as described in the Histology paragraph above. There were three male animals of 3–4 months of age originating from the same lines of breeding in each of the three experimental groups, WT, Acan-loxP^(+/+), and Acan-loxP^(+/−)Cre. At least five images from each of three sections from each brain of the individual animals were analyzed. Images were analyzed with an automated custom script using the EBImage package from R statistical software (Pau et al., 2010). For each experimental group, >150 and 50 PNN structures, respectively, were quantified by WFA and aggrecan histochemistry. Each PNN structure is represented as a dot in Figure 2, b and c. The three experimental groups were tested in one-way ANOVA with a Tukey's HSD post hoc test (Table 4).

Quantification of PV intensity by PV immunohistochemistry was conducted as described in the Histology section above. There were three male animals of 3–4 months of age originating from same lines of breeding, in each of the four experimental groups: WT, Acan-loxP^(+/+), Acan-loxP^(+/−)Cre, and Acan-loxP^(+/+)Cre. Five z-stack images, each consisting of 10 individual images 3 μm apart, were taken from each of five sections from each brain of the individual animals and analyzed. Images were analyzed with an automated custom script using the EBImage package from R statistical software (Pau et al., 2010). More than 150 PV⁺ cells were quantified for each experimental group and represented as a dot in Figure 2f. The four experimental groups were tested in one-way ANOVA with a Tukey's HSD post hoc test (Table 5).

Quantification of PNN intensity by WFA after local removal of Acan was conducted as described in the Histology section above. There were six male animals (3–4 months of age) originating from different litters but from the same breeding pair in which each of the two hemispheres contributed to two experimental groups: contralateral control and AAV9.Cre-injected hemispheres. Three sections from each of the individual animals were imaged. Each of the hemispheres of the individual animals is represented as a dot in Figure 3c. The two experimental groups were tested in a paired t test. Results are stated in the legend of Figure 3.

Assessing OD plasticity was conducted, as described in the Assessing OD plasticity section, in visual cortex using optical imaging of intrinsic signal section above. There were four, three, and seven male animals 5–7 months of age originating from the same line two breeding pairs and some from the same litters in the three different experimental groups, Acan-loxP^(+/+) untreated, Acan-loxP^(+/−) AAV9.Cre injected, and Acan-loxP^(+/+) AAV9.Cre injected. Responses to visual stimuli by each of the two eyes were recorded in four to six trials per eye. Each trial consisted of 20 repetitions of a visual stimulus of 120 s with simultaneous imaging of the cortex. Initially, baseline responses by the two individual eyes and corresponding ODI for all animals were determined (Fig. 1c–e, filled colored circles indicate population means, open gray circles indicate individual animals). After monocular deprivation (MD) of 4 d, the trials were repeated. Eye responses and ODI are represented in similar fashion. A shift in eye responses and ODI induced by MD of 4 d were tested in a paired t test within the three experimental groups or with Wilcoxon signed rank test whether the distribution was not normal. Results are stated in the legend of Figure 4.

Assessing memory of spontaneous object recognition was conducted as described in the Assessing memory by spontaneous object recognition behavioral task section above. There were 7, 9 (1 of which was singly housed), 10, and 7 male animals ∼3 months of age, respectively, originating from the same line of breeding in each of the four experimental groups, WT, Acan-loxP^(+/+), Acan-loxP^(+/−)Cre, and Acan-loxP^(+/+)Cre. Average exploration times for all experiments in each of the experimental groups varied between 40 and 50 s (Fig. 5b). To assess differences in the ability to retain memory (Fig. 5c), ANOVA was used to compare variances between the groups before engaging in post hoc analysis. A significant difference was found in the time point variable, which then justified the exploration of the time point differences within groups and reduced the likelihood that significant values were reached through chance (Table 6).