Together JUN and DDIT3 (CHOP) control retinal ganglion cell death after axonal injury

Mice were housed on a 12-h light dark cycle and fed chow and water ad libitum. Five different alleles were used to generate four different mice strains: 1) mice deficient in Jun (also known as cJun), 2) mice deficient in Ddit3, 3) mice deficient in both Jun and Ddit3, and 4) mice deficient in Dlk. To generate animals conditionally deficient in Jun, a floxed allele of Jun [44] and Six3-cre recombinase (a neural retina cre, Jackson Laboratory, Stock# 019755) were crossed to generate 1) animals carrying the recombined floxed alleles, referred to as Jun ^−/− or Jun deficient (Jun ^fl/flSix3cre⁺), 2) heterozygote animals referred to as Jun ^+/−(Jun ^+/flSix3cre⁺), and 3) animals carrying non-recombined floxed alleles or wildtype alleles with or without the cre recombinase referred to as Jun ^+/+ or wildtype (WT, Jun ^+/+ Six3cre ⁻ , Jun ^+/+ Six3cre ⁺ , Jun ^+/fl Six3cre ⁻ , or Jun ^fl/fl Six3cre ⁻ ). Mice carrying the null allele for Ddit3 were acquired from Jackson Laboratory (Stock# 005530) and intercrossed to generate: 1) animals carrying two copies of the null allele, referred to as Ddit3 ^−/− or Ddit3 deficient, 2) heterozygote animals carrying one copy of the null allele referred to as Ddit3 ^−/+, and 3) animals carrying no copies of the null allele, referred to as Ddit3 ^+/+ or WT in the text. Animals deficient in both Jun and Ddit3 were generated by crossing animals carrying the floxed allele of Jun and Six3-cre with animals carrying the null allele for Ddit3 to generate Ddit3^−/−;Jun ^fl/flSix3cre⁺ animals referred to as Jun/Ddit3 deficient animals in the text. To generate animals conditionally deficient in Dlk, a floxed allele of Dlk [45] and Six3-cre recombinase (a neural retina cre) were crossed on a C57BL/6 J background to generate 1) animals carrying the recombined floxed alleles, referred to as Dlk ^−/− or Dlk deficient (Dlk ^f/flSix3cre⁺) and 2) animals carrying non-recombined floxed alleles or wildtype alleles with or without the cre recombinase referred to as Dlk ^+/+ or WT (Dlk ^+/+ Six3cre ⁻ , Dlk ^+/+ Six3cre ⁺ , Dlk ^+/fl Six3cre ⁻ , or Dlk ^fl/fl Six3cre ⁻ ). The Jun ^fl and Dlk ^fl alleles were all backcrossed at least 3 times to the C57BL/6 J genetic background prior to intercrossing with the Six3cre allele that had been backcrossed at least 10 generations into the C57BL/6 J genetic background. The Ddit3 allele was obtained already backcrossed into C57BL/6 J for 5 generations. All experiments were conducted in adherence with the Association for Research in Vision and Ophthalmology’s statement on the use of animals in ophthalmic and vision research and were approved by the University of Rochester’s University Committee on Animal Resources.

Mice were anesthetized with 100 mg/kg ketamine and 10 mg/kg xylazine. Optic nerve crush (ONC) was completed as previously described [22, 23, 43, 46]. Briefly, the optic nerve was surgically exposed and crushed immediately behind the eye with a pair of self-closing forceps (Roboz RS-5027) for 5 s. Controls included contralateral eyes that had not been manipulated and eyes in which a sham surgery was performed, where the optic nerve was exposed but not crushed. Animals were harvested at 1, 5, 14, 35, 60, and 120 days following ONC or sham surgery.

Eyes to be processed for retinal morphology were fixed in 2.5% glutaraldehyde, 2% paraformaldehyde (PFA) for 24 h at 4 °C. After dehydration, eyes were embedded in Technovit (Electron Microscopy Services) and sectioned at 2.5 μm. Sections that included the optic nerve were stained with Multiple Stain Solution (Polysciences, Inc). Eyes to be processed for retinal flat mount immunohistochemistry were fixed in 4% PFA for two hours at room temperature. The anterior segment was removed and the posterior segment was processed for retinal flat mount immunohistochemistry as has been previously described [22, 23, 43, 46].

For retinal flat mount immunohistochemistry, retinas were dissected free of the optic cup and were blocked in 10% horse serum in 0.3% TritonX in 1 M PBS (phosphate buffered saline) for 24 h at 4 °C and then incubated in primary antibody for 72 h at 4 °C. Primary antibodies included: rabbit anti-JUN (Abcam, 1:250), rabbit anti-pJUN (Cell Signaling, 1:250) and mouse βIII tubulin (TUJ1; Covance, 1:1000). Whole retinas were washed in PBS and then incubated in secondary antibody, Alexafluor-conjugated secondary antibodies (Invitrogen), for 48 h at 4 °C before being washed in PBS and mounted RGC side up with Fluorogel in TRIS buffer (Electron Microscopy Sciences). Eight 40× fields were obtained for quantification of TUJ1+ and JUN+ cells. Images were taken 220 μm from the edge of the retina and equally spaced around the periphery of the retina (two fields per quadrant) as RGC density is known to vary across different dorso-ventral/medial-lateral retinal quadrants. Quantification was completed using the cell-counter tool in ImageJ.

Mouse retinas were dissected and placed in 100 μl ice cold lysis buffer (1X RIPA buffer (Santa Cruz 24,948) containing 50 μM sodium fluoride, 2 mM PMSF, 2 mM sodium orthovanadate, 1X protease/phosphatase inhibitor cocktail (Cell Signaling 5872S)). Tissue was lysed by sonication (Bransa Digital Sonifier, 10% amplitude for 3 s) prior to spinning down cellular debris in a microcentrifuge (10,000 rotations per minute, 4°C, 5 min). 10 μl of supernatant was combined with 10 μl 2X Laemmli loading buffer (Bio-Rad) and boiled for 10 min. Samples were fractionated by SDS-PAGE on a 12% gel and transferred to a PVDF membrane (transfer buffer: 1X Tris-glycine (Bio-Rad), 20% methanol). Membranes were rinsed with double-distilled H₂O and treated with a Qentix Western Blot Signal Enhancer kit (Thermo Scientific 21,050) according to manufacturer’s instructions. Membranes were then blocked at room temperature using 5% milk in TBST buffer for 1 h. Membranes were treated overnight at 4°C with one of the following primary antibodies: rabbit anti-JUN (Cell Signaling 9165S, 1:750) or mouse anti-GAPDH (Calbiochem CB1001, 1:2000). The following day membranes were washed 3X with TBST prior to treatment with secondary antibodies: HRP-conjugated anti-rabbit IgG (Bio-Rad 170–6515, 1:5000) or HRP-conjugated anti-mouse IgG (Bio-Rad 170–6516, 1:5000). Immunoreactive bands were detected using a chemiluminescence kit (Immun-star, Bio-Rad 170–5070) prior to exposure using either film or digital detection equipment (Azure Biosystems c500). Membranes were occasionally stripped following development and treated with another primary antibody (stripping buffer: 0.1 M Tris-Cl pH 6.8, 2% SDS, 0.7% β-mercaptoethanol). Densitometric analysis was conducted using ImageJ software, and pixel densities of experimental bands were normalized to those of GAPDH loading controls.

Optic nerve compound action potentials (CAPs) were recorded as described previously [43]. Briefly, animals were euthanized with CO₂, and optic nerves were dissected by transecting close to the eye and to the chiasm. Optic nerves were incubated for a minimum of 60 min in artificial cerebrospinal fluid (ACSF) aerated with 95% oxygen/5% CO₂ at room temperature. ACSF was prepared with (in mM): 125 NaCl, 1.25 NaH₂PO₄, 25 glucose, 25 NaHCO₃, 2.5 CaCl₂, 1.3 MgCl₂, and 2.5 KCl. The recording chamber was temperature controlled and perfused with oxygenated ACSF. The nerve was gently drawn into suction pipette electrodes containing ACSF at both the stimulating (retinal) end and the recording (chiasm) end. Stimuli consisted of supramaximal constant current pulses of 50 microseconds duration. CAP amplitudes were normalized by maintaining the ratio of the recording pipet plus nerve resistance to the recording pipet resistance at 1.7 [43]. All traces were taken at 25 °C. Data were digitized and then analyzed off-line.

To determine if Jun (also known as cJun) and/or Ddit3 (which encodes the protein CHOP) is necessary for retinal development; adult retinas deficient in Jun, Ddit3, and combined Jun/Ddit3 were evaluated. Jun was conditionally deleted in the retina using Six3-cre, an established cre recombinase which deletes floxed alleles in the early developing neuroepithelium [47]. Animals carrying the null allele for Ddit3 were evaluated independently and also crossed with animals carrying the Jun floxed allele and Six3-cre to generate animals deficient in both Jun and Ddit3. Deficiency in either Jun or Ddit3 or the combined deficiency of Jun and Ddit3 did not alter morphology of the retina (Fig. 1A). Furthermore, no differences were observed across subgroups in the number of TUJ1+ RGCs in adult animals (Fig. 1B; P > 0.05; n = 8 per group). Thus, neither Jun nor Ddit3 deficiency appears to alter the normal development of the retina. To assess the recombination efficiency of Six3-cre, the number of RGCs expressing JUN was evaluated one day after ONC. Consistent with previous studies, JUN accumulation was readily detected in the somas of RGCs of wildtype animals after axonal injury but was not easily detected in sham retinas (Fig. 1C), [22, 23, 48]. In retinas with conditional deletion of Jun, JUN accumulation after ONC was detected in approximately 22% of RGCs after ONC compared to wildtype retinas (Fig. 1D, P<0.001). Western blot analysis was completed to assess the level of JUN in wildtype and Jun deficient animals (Fig. E and F). In unmanipulated eyes, JUN was reduced 85% (p < 0.001) in Jun deficient retinas as compared to wildtype retinas. After ONC, JUN was reduced 92% (p = 0.004) in Jun deficient eyes as compared to wildtype animals. Thus, while recombination was not complete, Jun-mediated changes should be greatly reduced in Jun deficient conditional mutants.

Previously, JUN has been shown to be an important pro-apoptotic signaling pathway after glaucoma-relevant axon injury [21‐26, 28‐30, 35, 36]. DDIT3 has also been shown to be expressed after axonal insult [23, 31, 33, 34]. Furthermore, there is evidence to suggest that DDIT3 and JUN may act in the same signaling pathway. Accumulation of the DDIT3-dependent micro-RNA, miR-216b, has been found to directly target JUN [49]. Previously, Jun deficiency was found to not prevent the upregulation of DDIT3 protein after crush [23]. To determine if JUN is regulated by Ddit3, immunohistochemistry was performed on wildtype and Ddit3 deficient retinas. JUN accumulated in RGC somas after optic nerve injury (ONC) in both WT and Ddit3 deficient retinas (Fig. 2A). Quantification of JUN positive RGCs demonstrated no significant difference between JUN positive RGCs in wildtype and Ddit3 deficient retinas. Likewise, quantification of phosphorylated JUN positive RGCs demonstrated no significance difference between wildtype and Ddit3 deficient retinas after ONC (Fig. 2B). Western blot analysis was then completed both in unmanipulated eyes and after ONC to quantify the levels of JUN protein in wildtype and Ddit3 deficient retinas. No significant difference in JUN levels was observed between wildtype and Ddit3 deficient retinas in either condition. Together these results suggest that JUN expression is not dependent on Ddit3 and that Jun and Ddit3 may be independently regulated after optic nerve injury.

Since our results suggest that Jun and Ddit3 are independently upregulated after optic nerve injury, animals deficient in both Jun and Ddit3 were generated to determine if the protection afforded by the individual deficiencies is additive. RGC survival was assessed 14, 35, 60, and 120 days after ONC with immunohistochemistry in retinal flat mounts using the RGC specific marker, TUJ1. Consistent with previous reports, deficiency of either Jun or Ddit3 significantly increased RGC survival 14 and 35 days after ONC compared to WT animals [Fig. 3, Tables 1, 22, 33]. Jun/Ddit3 deficient animals also had increased survival 14 and 35 days after ONC compared to WT animals (14 days after ONC- WT: 29% survival, Ddit3: 51%, Jun: 93%, Jun/Ddit3: 100%; 35 days after ONC- WT: 22%, Ddit3: 48%, Jun: 75%, Jun/Ddit3: 89%; P < 0.001 for all comparisons against WT animals) Furthermore, combined Jun/Ddit3 deficient animals demonstrated increased survival compared to animals deficient in either Jun or Ddit3 alone at 35 days (p < 0.001). To determine if the protection afforded to RGCs by Jun/Ddit3 deficiency was sustained, as in in Bax deficient mice [7, 50], a cohort of animals was examined 60 and 120 days after ONC. At both of these time points, RGC survival in Jun, Ddit3, and Jun/Ddit3 deficient retinas was increased compared to WT retinas. RGC survival in combined Jun/Ddit3 deficient mice was also significantly increased compared to single deletion of Jun or Ddit3 in animals at both 60 days and 120 days after ONC (60 days after ONC- WT: 14 survival, Ddit3: 33%, Jun: 54%, Jun/Ddit3: 83%; 120 days after ONC- WT: 7%, Ddit3: 25%, Jun: 48%, Jun/Ddit3: 75%; P < 0.001 for all comparisons). The RGC protection conferred by single Jun deficiency and combined Jun and Ddit3 deficiency may be even higher since ~22% of RGCs in Jun and Jun/Ddit3 deficient retinas still express JUN due to incomplete recombination of the Jun floxed allele. Thus, the RGC protection afforded by combined Jun/Ddit3 deficiency is sustained after axonal injury. These results demonstrate that JUN and DDIT3 act as the two principal signaling nodes through which all pro-apoptotic signaling converges resulting in RGC death after axonal injury.

Dual leucine kinase (DLK) is a mitogen-activated protein kinase kinase kinase (MAP3K) and a critical regulator of RGC death after optic nerve injury [35, 36, 43]. Pharmacologic inhibition of DLK has been shown to be protective in rodent models of ocular hypertension suggesting the importance of this molecule in the pathogenesis of glaucoma [36]. After axonal injury, DLK is a major activator of JNK signaling and regulates both apoptosis and Wallerian degeneration. After optic nerve injury, Dlk deficiency decreases the somal pool of JNK, attenuates somal JUN accumulation, and ultimately increases survival of RGCs as compared to WT animals [35, 36, 43]. Though protective of RGC somas, Dlk deficiency does not prevent activation of axonal JNK [43]. Dlk has also been proposed to regulate Ddit3 expression after axonal injury. Unbiased gene expression profiling after mechanical axonal injury demonstrated that Dlk deficiency significantly reduced the expression of DDIT3 [35]. Together these results raise the possibility that DLK could be upstream of both Jun and Ddit3, thus making Dlk an important common upstream regulator of RGC death after axonal injury. To test this hypothesis, RGC survival was directly compared in WT, Dlk, and Jun/Ddit3 deficient retinas 35 days after ONC. This time point was chosen because there was significant difference in RGC loss between WT, Jun, Ddit3 and Jun/Ddit3 deficiency mice after ONC. TUJ1+ cell counts confirmed that both Dlk and Jun/Ddit3 deficiency provide significant protection to RGCs as compared to WT animals (Fig. 4; P < 0.001 n ≥ 7 per group; WT: 20.4% survival, Dlk: 64.3%, Jun/Ddit3: 88.5%). Furthermore, there was greater protection in Jun/Ddit3 deficient retinas compared to Dlk deficient retinas (p < 0.001) suggesting Dlk may not be the sole upstream regulator of JUN and DDIT3 activation.

RGC injury and axonal degeneration are important components of glaucomatous neurodegeneration [2, 8, 9]. After axonal injury, there are differential responses within the RGC somal and axonal compartments [51, 52]. There is spatial compartmentalization of molecular mechanisms that control RGC somal (cell body) and RGC axonal degeneration. Thus, it is necessary to test the role of JUN and DDIT3 in both somal death and axonal degeneration. DDIT3 deficiency promotes optic nerve survival in models of glaucoma, including mechanical optic nerve injury and increased intraocular pressure [33, 34, 37]. Jun deficiency has also been shown to protect axons from degeneration in sensory neuron culture [53]. In the DBA/2 J ocular hypertensive mouse model of glaucoma, Jun deficiency protected RGC somas but not axons from glaucomatous injury [54]. Thus, JUN does not appear to be required for axonal degeneration in adult RGCs after a glaucomatous injury. To test the hypothesis that single and/or combined deficiency of Jun and Ddit3 may protect against axonal degeneration, optic nerve function was tested after mechanical optic nerve injury. Compound action potentials (CAP) were measured in WT, Jun, Ddit3, and Jun/Ddit3 deficient mice five days after ONC, a time when WT animals demonstrate significant reduction in CAP amplitudes [43]. No differences in amplitudes were observed across the four groups, suggesting that deficiency of Jun and Ddit3, either singly or together, does not prevent for axonal degeneration after mechanical optic nerve injury (Fig. 5; P > 0.05 for all comparisons; n = 4 for all groups).