Effect of Glial Cell Line-Derived Neurotrophic Factor on Retinal Function after Experimental Branch Retinal Vein Occlusion

Abstract

Purpose.: The objective of the study was to investigate the effect of glial cell line–derived neurotrophic factor (GDNF) on the multifocal electroretinogram (mfERG) following an induced branch retinal vein occlusion (BRVO) in pigs.

Methods.: Electrophysiological examination of the retina was performed in 20 pigs with standard and four-frame mfERG 4 weeks after induced BRVO and intravitreal injection of GDNF or vehicle. BRVO was induced by intraocular diathermia of the superior retinal vein. Inner retinal function was measured by analysis of the four-frame mfERG (iN1) and outer retinal function with standard mfERG (P1).

Results.: In GDNF-treated BRVO eyes, P1 and iN1 amplitudes (P = 0.51 and 0.78) or implicit times (P = 0.08 and 0.99) did not differ from those in healthy fellow eyes. After vehicle injection, P1 and iN1 amplitudes of BRVO eyes were significantly lower than in the healthy fellow eye (P = 0.022 and 0.013). The log ratios of mfERG amplitudes between experimental and healthy fellow eyes were calculated (BRVO/healthy). GDNF improved the ratios of the four-frame mfERG (1.29 [0.88–1.88]) compared with vehicle (0.32 [0.21–0.50], P < 0.001). Equally, GDNF improved the ratios of the standard mfERG; GDNF (0.75 [0.51–1.10]) and vehicle (0.42 [0.27–0.63], P = 0.048).

Conclusions.: GDNF appears neuroprotective on retinal electrophysiological function after BRVO. The efficacy and safety of GDNF remain to be investigated in primate eyes.

Introduction

Branch retinal vein occlusion (BRVO) is a common retinal vascular disease in which interruption of venous blood flow almost always occurs at arteriovenous crossings, ^1,2 causing decreased capillary blood flow. ^3,4 Typical findings include segmental intraretinal hemorrhages, vitreal hemorrhages, and macular ischemia. Hypoxia-induced cytokines, such as vascular endothelial growth factor (VEGF), and venous congestion may lead to macular edema and neovascularization. ⁵ Amplitudes as well as implicit times of the multifocal electroretinogram (mfERG) are affected by BRVO, ⁶ and correlate with angiographic signs of macular ischemia. ⁷ Amplitudes correlate well with visual field findings ⁸ and visual acuity. ⁷  

A functional model of experimental BRVO in pigs was presented previously, with detection of focal changes in retinal electrophysiological function with mfERG. ⁹ Localized retinal damage caused by BRVO led to decreased amplitudes and prolonged implicit times of the standard and induced mfERG, representing outer ¹⁰ and inner retinal function, ^11–16 respectively. 

As glial cell line–derived neurotrophic factor (GDNF) and its receptors are synthesized in the retina, ^17,18 the factor may have an innate neurotrophic role in the retina. GDNF has been proposed as a neuroprotectant for retinal ganglion cells (RGCs) in axotomized animal eyes, ^19–23 while other studies have shown that GDNF increases the survival of RGCs in glaucomatous retinas of rats ²⁴ and mice, ²⁵ as well as in postischemic retinas of pigs. ²⁶ Moreover, GDNF maintained photoreceptors following retinal degeneration in mice, ²⁷ in rats, ²⁸ and in rat eyes with detached retinas. ²⁹ Recently, GDNF was proven to support transplanted sheets of fetal retina and improve visual function by reinforcing proper synaptic connections and reducing apoptosis. ³⁰ Thus protection of both inner and outer retinal function was expected. 

In the present study, rescue of retinal function was attempted with GDNF, and the effect was evaluated with mfERG. 

Methods

Animals

All experiments were performed in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and The Danish Animal Experiments Inspectorate granted permission for the use of the animals (permission no. 2007/561-1386). Animals were tranquilized, anesthetized, and ventilated as previously described. ⁹ In 34 female domestic pigs of Danish Landrace (Duroc/Hampshire/Yorkshire) breed (age: 3–4 months; weight: 24–31 kg), right eyes underwent surgery to induce BRVO. 

Surgery

Right eyes were anesthetized, dilated, and disinfected as previously described. ⁹ Following temporal canthotomy and conjunctival incision, three sclerotomies were performed 2 mm posterior to the corneal limbus. An infusion line was secured inferiorly with balanced salt solution (BSS Plus; Alcon, Rødovre, Denmark), and a modified blunt bipolar diathermy probe and a light source (Accurus Surgical System; Alcon) were inserted temporally and nasally, respectively. The diathermy probe was placed over the superior vein one disc diameter superior to the disc and diathermia was initiated, lasting 6 seconds at 100% energy (7.50 W at 100 Ω). Sclera and conjunctiva were sutured with 7-0 Vicryl (Ethicon Inc., Norderstedt, Germany) after application of 1% chloramphenicol (Kloramfenikol “DAK”; Nycomed, Roskilde, Denmark). An injection of 100 ng GDNF (0.2 mL) or vehicle (0.2 mL PBS with 1% BSA) was administered in the midvitreous of BRVO eyes via a 30-guage needle. Afterward, bimanual palpation and indirect ophthalmoscopy were performed to exclude complications, and topical chloramphenicol ointment was given. Due to the intravitreal pharmacokinetics of GDNF, ³¹ the factor had to be injected twice to be effective for 4 weeks. Two weeks postoperatively, animals were tranquilized, and BRVO eyes were reinjected with GDNF or vehicle. Animals were randomized for GDNF or vehicle injection, and the investigator was blinded. 

Follow-up

Four weeks postsurgery, reliable porcine mfERGs could be obtained. ⁹ Animals were reanesthetized intravenously with the addition of a neuromuscular blocker to avoid eye movement, 2 mg pancuronium (DeltaSelect GmbH, Pfullingen, Germany), before each mfERG recording. Multifocal ERG measurements were conducted in an electrically shielded room under standardized lighting conditions (28 cd/m²), and dilated eyes were light adapted for 15 minutes. A Burian-Allen bipolar contact lens electrode (VERIS Infrared Illuminating Electrode; EDI, Inc., Redwood, CA) was placed on the cornea with a gel (Viscotears; Novartis, Copenhagen, Denmark) as contact fluid, and a reference electrode was placed behind the ear. Animals and all electrical equipment were electrically grounded. The mfERG equipment allowed continuous infrared (IR) fundus monitoring during recordings to assist in detection of eye movement. For each eye, recordings centered on the visual streak were obtained: first a standard one-frame mfERG, and secondly an induced or four-frame mfERG. ⁹ Recordings were made serially, and the order of testing (right or left eye first) was counterbalanced in both the GDNF and vehicle groups by block randomization. 

Finally, indirect ophthalmoscopy and bilateral color fundus photography were performed. Thereafter, pigs were euthanized by a lethal injection of 4 g pentobarbital with 400 mg lidocaine hydrochloride (Veterinærapoteket Københavns Universitet, Copenhagen, Denmark). All eyes were enucleated and fixed in 4% buffered formaldehyde, dissected, and embedded in paraffin. Vertical sections through the visual streak (Fig. 1) were cut at 4 μm, mounted on glass slides, and stained with hematoxylin and eosin (HE) for histopathological examination. Animals with vitreous opacities or sustained damage to the lens were discarded to prevent interference with the mfERG. Eleven GDNF and nine vehicle animals were examined. 

Figure 1. 

 

Fundus pictures and corresponding histological sections of experimental BRVO animal eyes treated with vehicle (A–C) and GDNF (D–F). In both BRVO eyes (A, D), the white scarring from the diathermia is present over the superior retinal vein 4 weeks after surgery. In addition, pigment dispersion is present in the BRVO area. These findings were more pronounced in vehicle-treated eyes (A) than in GDNF-treated eyes (D). Histological sections stained with HE confirm that vehicle-treated BRVO eyes show severe structural damage to all retinal layers including photoreceptors and numerous inflammatory cells in the superior visual streak retina (B). This is in contrast to the GDNF-treated BRVO eye, which displays mild inflammation, mild edema of the inner retina, and thinning of the photoreceptor layer (E). Inferior visual streak (outside the BRVO area) of the vehicle (C)- and GDNF (F)-treated eyes shows similar retinal morphology. Bar equals 50 μm. ILM, inner limiting membrane; NFL, nerve fiber layer; GCL, ganglion cell layer; IPL, inner plexiforme layer; INL, inner nuclear layer; OPL, outer plexiforme layer; ONL, outer nuclear layer; ELM, external limiting membrane; PRL, photoreceptor layer; RPE, retinal pigment epithelium.

Multifocal Electroretinography Settings

Multifocal electroretinograms were recorded using a VERIS Multifocal System with VERIS 6.0.8 software (EDI, Inc.). The standard mfERG stimulation consisted of one frame that underwent a pseudorandom m-sequence of flash or dark frames, whereas the induced mfERG stimulation consisted of four frames: one m-sequence frame, one dark frame, one flash frame, and another dark frame. Both modalities have been previously described in connection with the BRVO model. ⁹ Briefly, the standard mfERG records the response to a focal light stimulus and is primarily bipolar and photoreceptor driven. ¹⁰ In contrast, the induced (or four-frame) mfERG records the adaptive response due to the interactions of a focal and global flash, and the cellular origin is the inner retina (amacrine and ganglion cells). ^11–16 A black-and-white 103 unscaled hexagon stimulus pattern at a frame rate of 75 Hz, with 16 samples per frame, was used. The m-exponent was 15, and the durations of the recordings were 7.17 and 14.37 minutes, respectively. Signals were band-pass filtered outside 10 to 300 Hz, and neither bipolar artifact rejecter nor spatial averaging was used. Gain was 20 K. 

Multifocal Electroretinography Analysis

The IR fundus pictures from the VERIS system were aligned with corresponding fundus pictures to determine the position of the individual mfERG responses on the retina. A horizontal raphe divides the fundus into two areas supplied by the superior and inferior vasculature, respectively. Induced BRVO affects the superior area, including the superior half of the visual streak. The inferior area contains both visual streak and nonvisual streak and the optic disc; this anatomical division corresponds with photoreceptor density. ³² Responses obtained entirely within the BRVO-affected superior visual streak (area 1) and within the inferior visual streak (area 2) were summed. Corresponding areas of healthy fellow eyes were used as reference to reduce variation due to anesthesia and interindividual variation. To obtain normally distributed amplitudes and stabilize the variance, the log-transformed amplitudes were analyzed as previously described. ⁹ Peak P1 of the standard MfERG and through iN1 of the induced mfERG were analyzed statistically as reported in a previous study. ⁹ Mean and 95% confidence interval (mean CI) of retransformed amplitudes and implicit times are presented. Data were considered continuous, and Student's t-test and two-way ANOVA were used with Holm-Sidak correction for multiple comparisons. Statistics and graphs were made in SigmaStat/SigmaPlot (Systat Software Inc., San Jose, CA). 

Results

Effects of GDNF on BRVO

Tables 1 and 2 show the standard and induced mfERG data from BRVO and healthy eyes 4 weeks after induced BRVO. 

Table 1. 

 

Amplitudes and Implicit Times of the Major Peak P1 of the Standard mfERG

		BRVO Eye	Healthy Eye	BRVO/Healthy
P1, vehicle
Amplitude (nV/deg2), mean (CI)	Area 1	11.6 (5.90–22.7)	27.8 (22.9–33.7)	0.42 (0.27–0.63)
	Area 2	15.2 (9.97–23.2)	23.3 (19.3–28.1)	0.65 (0.43–0.99)
Implicit time (ms), mean (CI)	Area 1	24.1 (23.6–24.6)	23.2 (22.8–23.7)	1.04 (1.02–1.05)
	Area 2	24.4 (23.9–24.8)	23.7 (23.4–24.0)	1.03 (1.01–1.04)
P1, GDNF
Amplitude (nV/deg2), mean (CI)	Area 1	20.9 (13.2–33.1)	27.8 (24.3–31,8)	0.75 (0.51–1.10)
	Area 2	21.4 (16.4–27.9)	24.4 (21.1–28.2)	0.88 (0.60–1.28)
Implicit time (ms), mean (CI)	Area 1	23.9 (23.6–24.3)	23.4 (23.0–23.8)	1.02 (1.01–1.04)
	Area 2	24.1 (23.7–24.5)	23.9 (23.3–24.4)	1.01 (1.00–1.03)

 Values from BRVO eyes and healthy fellow eyes from both treatment groups (GDNF or vehicle) are given as well as the ratio between the eyes (BRVO/healthy). Mean and 95% confidence interval (mean CI) of retransformed log amplitudes and implicit times are given. N = 11 for GDNF and n = 9 for vehicle.

Table 2. 

 

Amplitudes and Implicit Times of the Major Deflection iN1 of the Four-Frame mfERG

		BRVO Eye	Healthy Eye	BRVO/Healthy
iN1, vehicle
Amplitude (nV/deg2), mean (CI)	Area 1	12.8 (6.11–26.9)	39.5 (30.2–49.9)	0.32 (0.21–0.50)
	Area 2	20.1 (13.2–30.4)	37.5 (31.8–44.2)	0.53 (0.34–0,83)
Implicit time (ms), mean (CI)	Area 1	52.8 (52.1–53.5)	52.6 (51.8–53.4)	1.00 (0.99–1.02)
	Area 2	52.7 (52.1–53.3)	53.0 (52.4–53.6)	0.99 (0.98–1.01)
iN1, GDNF
Amplitude (nV/deg2), mean (CI)	Area 1	27.3 (17.8–42.1)	21.2 (14.8–30.4)	1.29 (0.88–1.88)
	Area 2	30.6 (22.7–41.2)	23.0 (17.6–30.0)	1.33 (0.91–1.94)
Implicit time (ms), mean (CI)	Area 1	52.7 (52.2–53.2)	52.3 (51.3–53.3)	1.01 (0.99–1.02)
	Area 2	52,6 (52.0–53.2)	52.4 (51.6–53.3)	1.00 (0.99–1.02)

 Values from BRVO eyes and healthy fellow eyes from both treatment groups (GDNF or vehicle) are given as well as the ratio between the eyes (BRVO/healthy). Mean and 95% confidence interval (mean CI) of retransformed log amplitudes and implicit times are given. N = 11 for GDNF and n = 9 for vehicle.

In the GDNF-treated BRVO eyes, neither the standard nor the induced mfERGs were different from those of healthy fellow eyes: In the BRVO-affected superior visual streak (area 1, Fig. 2), P1 and iN1 amplitudes (P = 0.51 and 0.78, respectively) and implicit times (P = 0.08 and 0.99, respectively) did not differ from those in healthy fellow eyes. Similarly, no difference existed in amplitudes (P = 0.99 for P1 and P = 0.18 for iN1) or implicit times (P = 0.99 for P1 and iN1) in the inferior visual streak (area 2, Fig. 2). 

Figure 2. 

 

Multifocal ERG recordings from BRVO and healthy eyes of GDNF- and vehicle-treated animals. (A) Multifocal ERG traces merged with fundus photography illustrate the position of the mfERG responses on the fundus. (B) The 103 standard mfERG traces obtained from each eye. Note the smaller responses corresponding to the BRVO area of superior visual streak in the vehicle-treated eye. Responses obtained outside the visual streak (inferiorly and over the optic nerve) are smaller than those from visual streak in healthy eyes and the GDNF-treated eye. (C) The grouping pattern of the mfERG hexagons. Area 1, superior to the horizontal raphe, is directly affected by the induced BRVO whereas the inferior visual streak (area 2) is not. (D) Summed standard mfERG responses from area 1 (top) and area 2 (bottom) as presented by the VERIS software. The major peak P1 is labeled. (E) Summed four-frame mfERG responses from area 1 (top) and area 2 (bottom) as presented by the VERIS software. The major negative deflection iN1 is labeled.

In addition, the effect of GDNF in BRVO eyes was evaluated through comparison of the ratios of amplitudes and implicit times (BRVO/healthy) between the GDNF and vehicle groups. At any given time, we evaluated the ratio of the log-transformed mfERG amplitudes and implicit times from the BRVO eye normalized to that of the healthy eye. This minimized any interindividual anesthesia- and time-dependent variability in the data. ⁹  

A two-way ANOVA showed that GDNF-treated eyes displayed higher iN1 amplitudes than vehicle-treated eyes in both areas 1 and 2 (Fig. 3, P < 0.001 and 0.004, respectively). No significant effect of area (P = 0.21) or GDNF × area (P = 0.27) on iN1 amplitudes was found. The corresponding ratios of the P1 amplitudes were significantly larger in the GDNF than in the vehicle group in area 1 but not in area 2 (Fig. 3, P = 0.048 and 0.31, respectively). No effect on P1 amplitudes of area (P = 0.15) or GDNF × area (P = 0.48) could be detected. 

Figure 3. 

 

Amplitude ratios between BRVO and healthy eyes (BRVO/healthy) for both vehicle- and GDNF-treated animals. Retransformed mean and 95% confidence intervals are shown. All ratios from vehicle-treated eyes were below one, indicating significantly lower amplitudes in the BRVO eyes than in healthy fellow eyes. In GDNF-treated eyes, none of the ratios were significantly different from one, indicating no difference in amplitudes between BRVO and healthy fellow eyes. The iN1 amplitudes displayed significantly higher ratios in the GDNF-treated eyes than in vehicle-treated eyes whereas this was not the case for P1 amplitudes.

No difference in the ratios (BRVO/healthy) was found between treatment groups for P1 or iN1 implicit times (P > 0.05 for any area) with ANOVA. 

Consequences of BRVO

To assess the electrophysiological consequences of BRVO, the vehicle-treated eyes were analyzed. For the BRVO-affected superior visual streak (area 1), the P1 and iN1 amplitudes were significantly lower in the BRVO eye than in the healthy fellow eye (Tables 1 and 2, P = 0.022 and 0.013, respectively). Accordingly, the ratios of the P1 and iN1 amplitudes between the BRVO eye and the healthy fellow eye (BRVO/healthy) were less than one (Fig. 3). Implicit times of the standard mfERG (P1) were prolonged significantly after surgery in area 1 (Table 1, P = 0.049), but this was not the case for the induced mfERG (Table 2, P = 0.70). 

The mfERGs of the inferior visual streak (area 2) were less affected by the BRVO: The P1 and iN1 amplitude ratios (BRVO/healthy) obtained in area 2 appeared larger than the ratios obtained in area 1, as seen from Figure 3. However, the difference was not statistically significant (P = 0.21 and 0.26, respectively). In area 2, the iN1 amplitudes were significantly lower in the BRVO eye than in the healthy fellow eye (P = 0.032) whereas this was not observed for P1 (P = 0.36). Implicit times from area 2 were not statistically significantly different between eyes (P = 0.093 and 0.99 for P1 and iN1, respectively). 

Clinical and Histopathological Findings

Postoperatively, fundoscopy showed complete occlusion of the superior retinal vein in all BRVO eyes. This initial occlusion was characterized by venous dilation and tortuosity peripheral to the burn, and intraretinal hemorrhages were observed. All operated eyes displayed very homologous pathology. 

Four weeks after surgery, ophthalmoscopy and color fundus photography (Fig. 1) revealed signs of pigment abnormalities and venous sheathing in the BRVO area, and in all eyes the retinal scar from the diathermia was visible. However, fundoscopy showed great variability in fundus appearances within both the GDNF and vehicle groups. The histopathological sections confirmed a marked difference between BRVO-affected superior visual streak (area 1, Figs. 1B, 1E) and inferior visual streak (area 2, Figs. 1C, 1F). In BRVO-affected areas, variable degrees of cell loss, gliosis, and edema were observed in both groups, thus making quantification unconvincing. The structural damage to all the retinal layers including the photoreceptors was more prominent in the vehicle-treated BRVO eyes (Fig. 1B) than in the GDNF-treated BRVO eyes (Fig. 1E). The histopathological changes after BRVO were seen with less severe degeneration, edema, and destruction of the retinal layers after GDNF. 

Discussion

Consequences of GDNF Treatment in BRVO Eyes

The major endpoint of this study was the effect of GDNF on retinal electrophysiological function after BRVO. The data presented here support the findings of a neuroprotective effect of GDNF on the electrophysiological function of ischemia-reperfusion–damaged rat ³³ and porcine retinas. ²⁶ The effect was most pronounced on the induced mfERG, as iN1 amplitude ratios (BRVO/healthy) were greatest in the GDNF group and as GDNF-treated BRVO eyes produced amplitudes not significantly different from those of healthy eyes. Previously, after inner retinal damage, a strong correlation was found between histopathology and iN1 amplitudes in pigs, ¹⁵ and iN1 was nearly abolished by blocking retinal ganglion cell activity with tetrodotoxin. ¹¹ Hence, the ameliorative effect of GDNF on the iN1 amplitude suggests protection of neurons of the inner retina and their function. ^19–26,33 The induced mfERG records total activity in the innermost part of the retina, including ganglion and amacrine cells, but does not allow for analysis of individual cell types. ^11–16  

The standard mfERG data provide similar results. GDNF treatment resulted in P1 amplitudes of BRVO eyes that were not statistically lower than in healthy eyes (Fig. 3). This was in contrast to the untreated BRVO eyes, which displayed lower amplitudes than healthy eyes. In addition, the amplitude ratios (BRVO/healthy) of BRVO-affected area 1 were significantly higher in the GDNF group than in the vehicle group. Finally, implicit times were delayed in the vehicle group but not in area 2 of the GDNF group. Therefore, outer retinal function might be protected by GDNF. The P1 amplitude is predominantly of bipolar cell origin but is dependent on photoreceptor activity. ¹⁰ Thus our findings are consistent with previously reported anatomical protection of photoreceptors. ^27,29,34 This effect of GDNF on photoreceptors is likely an indirect effect mediated by retinal Müller cells. ³⁵ However, following ischemic retinal damage, GDNF did not improve ERG responses ³⁶ whereas improvement was seen after light-induced retinal degeneration. ³⁴  

Our results are important, as they present the functional consequences of GDNF treatment of retinal damage from a randomized and blinded study in a large animal. An interesting observation is that zero retinal activity does not completely abolish the mfERG response in pigs. Indeed, complete local destruction of the retina results in an amplitude ratio of approximately 50%, ³⁷ partly due to scattering of light within the eye. ³⁸ Figure 3 clearly demonstrates that ratios obtained from GDNF eyes were all above 0.5 whereas those of vehicle-treated eyes were not; some of the vehicle-treated BRVO eyes had close to zero retinal activity as opposed to none of the GDNF-treated eyes. 

We found a difference in the iN1 amplitudes between the untreated healthy eyes in the two groups (Table 2). This was most likely due to differences in the effects of anesthesia, as no efferent cerebral inhibition of the retina has been described; however, a systemic effect of GDNF cannot be ruled out. Variation due to anesthesia was minimized by intraindividual comparisons. 

Although quantification of histiological findings was not achieved, the observed preservation of retinal layering by GDNF corresponds with the electrophysiological findings. GDNF possibly supports retinal structure and proper synaptic connections, as has been suggested in transplanted sheets of fetal retina. ³⁰  

The BRVO Model

A smaller diathermia probe was used in the present study, but findings conformed to previous descriptions of pathology in this model. ⁹ The superior half of the visual streak was affected by the BRVO, but the inferior visual streak was not unaffected. These changes were probably brought about by hypoxia and decreased nitric oxide production, ³⁹ apoptosis, ⁴⁰ and general inflammation with alterations in retinal gene expression and Müller cell activity throughout the retina. ⁴¹ As expected with retinal ischemia, the inner retinal responses (iN1) were more severely depressed by the BRVO than outer retinal responses (P1). 

Future Perspectives

Induced BRVO displays spontaneous remission, as the occluded vein will reperfuse within 2 weeks. ⁹ This is probably a prerequisite in order for GDNF to attenuate the functional retinal damage. The ameliorative effect on the ischemic insult by GDNF is probably through the inhibition of apoptosis. ^21,42 Hence, suffering cells are more likely to survive until normal perfusion is restored. As such, this effect could be expected to improve the visual outcome of patients with retinal vein occlusions in whom retinocilliary collaterals have developed, ^43,44 in whom the thrombus has recanalized, ⁴⁵ and in whom a functional retinochoroidal anastemosis is successfully induced. ⁴⁶ In addition, other conditions with acute retinal damage (e.g., acute angle closure glaucoma and retinal detachment) might potentially be targeted by GDNF. 

In conclusion, GDNF appears neuroprotective toward retinal electrophysiological function after BRVO. The efficacy and safety of GDNF remain to be investigated in primate eyes. 

Footnotes

 Supported by The Michaelsen Foundation, The Danish Eye Health Society, The A.P. Møller Foundation for the Advancement of Medical Science, and The Velux Foundation.

Footnotes

 Disclosure: R. Ejstrup, None; M. la Cour, None; M.V. Kyhn, None; S. Heegaard, None; J.F. Kiilgaard, None