• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Neurosci. Author manuscript; available in PMC Oct 2, 2009.
Published in final edited form as:
PMCID: PMC2756297

BARHL2 differentially regulates the development of retinal amacrine and ganglion neurons


Through transcriptional regulations the BarH family of homeodomain proteins play essential roles in cell fate specification, cell differentiation, migration and survival. Barhl2, a member of the Barh gene family, is expressed in retinal ganglion cells (RGCs), amacrine cells (ACs) and horizontal cells. Here, to investigate the role of Barhl2 in retinal development, Barhl2 deficient mice were generated. Analysis of AC subtypes in Barhl2 deficient retinas suggests that Barhl2 plays a critical role in AC subtype determination. A significant reduction of glycinergic and GABAergic ACs with a substantial increase in the number of cholinergic ACs was observed in Barhl2-null retinas. Barhl2 is also critical for the development of a normal complement of RGCs. Barhl2 deficiency resulted in a 35% increase in RGCs undergoing apoptosis during development. Genetic analysis revealed that Barhl2 functions downstream of the Atoh7-Pou4f3 regulatory pathway and regulates the maturation and/or survival of RGCs. Thus, BARHL2 appears to have numerous roles in retinal development, including regulating neuronal subtype specification, differentiation, and survival.

Keywords: BarH, Math5, Atoh7, Pou4f2, Isl1, retinal ganglion cells, amacrine cells, retinal development, transcription factor


In retina, visual stimuli are received by photoreceptor cells and transferred via bipolar cells to retinal ganglion cells (RGCs). The two types of interneurons, amacrine and horizontal cells, function to integrate and modulate visual signal processing within retina. Based on their unique morphological, physiological, neurotransmitter, and connectivity properties, retinal neurons are further divided into subtypes. For instance, amacrine cells (ACs), which reside in the amacrine cell layer (ACL), an inner layer of the inner nuclear layer (INL), and the ganglion cell layer (GCL), synapse with bipolar cells and RGCs. They constitute the most diverse cell type within the retina with greater than 26 different subtypes based on their sub-laminar localization (e.g., displaced ACs), morphology (e.g., starburst, parasol and midget), and neurotransmitter types (e.g., GABAergic, dopaminergic, and cholinergic) (Masland, 2001b, a). Similarly, RGCs, the output neurons of the retina, are divided into more than ten different subtypes based on their morphology, physiological properties, and stratification patterns (Kong et al., 2005). While we have made significant progresses understanding the molecules important in generating the broad classes of retinal cell types, currently, the molecules defining retinal subtypes are poorly understood.

Previous studies have demonstrated that many transcription factors (TFs) play important roles in retinogenesis. During the development of RGCs, ATOH7 (MATH5), a basic helix-loop-helix (bHLH) TF, endows retinal precursors with the competency to form RGCs (Yang et al., 2003). In Atoh7-null retinas, more than 95% RGCs fail to form (Brown et al., 2001; Wang et al., 2001). POU4F2, a POU-homeodomain (HD) TF, and ISL1, a LIM-HD TF, are synergistically required for RGC differentiation and survival (Gan et al., 1999; Pan et al., 2008). Nevertheless, the loss of RGCs in mice lacking Pou4f2 or Atoh7 is not selective to any particular subtype (Lin et al., 2004), suggesting that they are general factors in RGC development. Among the key TFs in amacrine development, NEUROD4 and NEUROD1, both bHLH TFs, are redundantly required for amacrine genesis (Marquardt et al., 2001; Inoue et al., 2002) whereas retina-specific deletion of Pax6 HD TF results in the formation of only ACs (Marquardt et al., 2001). Mice lacking the winged helix/forkhead TF FOXN4 have fewer ACs and no horizontal cells (Li et al., 2004). However, the above TFs regulate AC generation in a non-subtype-specific manner. Recently studies show that BHLHB5, a bHLH TF, and ISL1 are expressed in GABAergic and cholinergic ACs, respectively and are necessary for their differentiation (Feng et al., 2006; Elshatory et al., 2007), suggesting that unique TFs determine amacrine subtype development.

Bar-class HD (BarH) proteins are evolutionarily conserved and play essential roles in organogenesis. In Drosophila, BarH1 and BarH2 are expressed in R1/6 photoreceptor cells and are required for their differentiations (Higashijima et al., 1992). The Xenopus BarH orthologue, Xbh1, promotes RGC differentiation (Patterson et al., 2000; Poggi et al., 2004). In mouse retina, BARHL2 expression is found in RGCs, ACs and horizontal cells. Forced expression of BARHL2 promotes the differentiation of glycinergic ACs at the expense of bipolar and Müller cells (Mo et al., 2004). Thus, BARHL2 appears to be an important molecule in mammalian retinal development. To understand the role of BARHL2 in retinal development, we generated a non-functional allele of Barhl2 in mice. In Barhl2-null retinas there was a selective loss of RGCs at late embryonic stages. Moreover, there was an overall loss of ACs that included a significant loss of both glycinergic and GABAergic in Barhl2-null retinas. Interestingly, the number of cholinergic ACs was greatly increased. These data demonstrate that BARHL2 is an important factor for the maintenance of RGCs and ACs as well as being critical for subtype specification of retinal neurons.

Materials and Methods

Generation and genotyping of Barhl2-lacZ and Barhl2-Cre mice

To generate Barhl2-null mice, we used Barhl2 coding sequences as a probe to isolate genomic sequences from mouse 129S6 BAC genomic library (CHORI, CA). The Barhl2-lacZ reporter and Barhl2-Cre knock-in construct were generated by inserting 3.5 kb Barhl2 5′ flanking sequences ending at the 20th nucleotide upstream of the initiation codon ATG and 3.5 kb 3′ sequence starting from the 68th nucleotide after ATG into 5′ and3′ multiple cloning sites of pKI-lacZ and pKII-Cre vectors (L.G., unpublished) at the HindIII and EcoRI sites, respectively. The insertion placed lacZ or Cre under the control of endogenous Barhl2 regulatory sequences. To generate Barhl2-null mice, AscI-linearized Barhl2-lacZ and Barhl2-Cre targeting constructs were electroporated into W4 mouse embryonic stem (ES) cells (Auerbach et al., 2000) and positive targeted ES cells were obtained by drug selection with G418 and FIAU. The targeted ES cells were confirmed by Southern blotting genotyping and were injected into C57BL/6J blastocysts to generate mouse chimeras. Barhl2-lacZ and Barhl2-Cre mice were generated in 129S6 and C57BL/6J mixed background as previously described (Gan et al., 1996; Gan et al., 1999). PCR was performed to genotype mice from heterozygous mating. PCR primers used to identify the wild type (WT) Barhl2 allele are 5′-ATGACAGCACTGGAAGGGGCC-3′ and 5′-ACGCTGGTGCTGTAGGGTGCACA-3′; the lacZ allele 5′-AGGGCCGCAAAACTATCC-3′ and 5′-ACTTCGGCACCTTACGCTTCTTCT-3′; and the Cre allele 5′-GCAGTGTCAACGCTTTTTAGTGTC-3′ and 5′-CCATGAGTGAACGAACCTGGTCG-3′. The Z/EG conditional GFP reporter mice were purchased from the Jackson Laboratory (Stock Number: 003920), and genotyping was performed according to protocols provided by Jackson Laboratory. Embryos were identified as E0.5 at noon on the day at which vaginal plugs were first observed. All animal procedures in this study were approved by University Committee of Animal Resources (UCAR) at University of Rochester.

Immunohistochemistry, in situ hybridization and X-Gal staining

Staged mouse embryos and enucleated eyes of postnatal mice were fixed with 4% paraformaldehyde in PBS for 1–2 hours at 4°C. Following the fixation, samples were cryopreserved with 20% sucrose, embedded in OCT compound, and cryosectioned at a thickness of 14 μm. Immunostaining, in situ hybridization, and X-Gal staining were performed as described (Pan et al., 2005). For whole mount immunostaining, animals were anesthetized with halothane and perfused with 4% paraformaldehyde in PBS. Eyecups were isolated and blocked in 10% horse serum with 0.5% Triton-X100 for four hours and incubated with primary antibody for 2–3 days at 4°C. Then, eyecups were washed in PBS and incubated with fluorescently labeled secondary antibody for one day at 4°C. After three rinses of 15 minutes each, retinas were dissected out and mounted in PBS on slides. Retinal images were captured with a Zeiss LSM 510 META confocal microscope. To quantify the numbers of immunolabeled cells, five or more retinas for each genotype were analyzed with Image J software (NIH).

The following antibodies and dilutions were used: rabbit anti-β-galactosidase (lacZ) (1:500, Chemicon), goat anti-POU4F2 (1:200, Santa Cruz), mouse anti-calbindin (1:2,000, Sigma), mouse anti-POU4F1 (1:200, Chemicon), mouse anti-syntaxin (1:5000, Santa Cruz), mouse anti-PAX6 (1:200, Developmental Studies Hybridoma Bank), mouse anti-Isl1/2 (1:200, DSHB), goat anti-ChAT (1:100, Chemicon), sheep anti-Chx10 (Exalpha), rabbit anti-PKCα (1:5,000, Sigma), mouse anti-GAD65 (1:200, BD Bioscience), rabbit anti-activated caspase3 (1:200, BD Pharmingen), goat anti-GLYT1 (1:5,000, Chemicon), rabbit anti-opsin (1:200), chicken anti-GFP (1:500, Abcam). Alexa-conjugated secondary antibodies (Molecular Probes) were used at a concentration of 1:500.

Multielectrode Array (MEA) Recording

The procedures of retina preparation, action potential recording and data analysis have been described previously in detail (Tian and Copenhagen, 2003). Briefly, mice were dark adapted for 30 minutes before euthanasia. Retinas were isolated under infrared illumination in oxygenated extracellular solution, which contained (in mM) NaCl 124, KCl 2.5, CaCl2 2, MgCl2 2, NaH2PO2 1.25, NaHCO3 26 and glucose 22 (pH 7.35 with 95% O2 and 5% CO2). Isolated retina was mounted in the MEA-60 chamber with the ganglion cell layer facing the recording electrodes and continuously perfused with oxygenated extracellular solution at 34°C. For spontaneous retinal waves, data were collected for 30 min. For light-evoked action potentials, a white rectangular moving bar (600 μm × 4000 μm) was used to stimulate the retinas, which moving upon black background and perpendicularly to its long axis at 1000 mm/sec and in 12 directions at 30° intervals with a pseudorandom sequence. The stimulations were repeated 30 times with random sequences. Data were collected using a PC-based interface card and software (Multi Channel System MCS GmbH, Reutlingen, Germany). The signals were filtered between 100 Hz (low cut off) and 200 or 250 kHz (high cut off). Offline data analysis was carried out on a PC computer using Offline Sorter (Plexon Inc, Dallas, TX).

MEA Data Analysis

For spontaneous retinal waves, the cell with highest firing rate was selected for each electrode for the calculation of correlation index. The correlation index for a given pair of neurons was calculated based on the following equation (Wong et al., 1993; Torborg et al., 2005).


Where NAB is the number of spike pairs from cell A and B for which cell B fires within ±100 ms of cell A, T is the total recording time, NA(0,T) is the total number of spikes in cell A and NB(0,T) is the total number of spikes in cell B. The frequencies of action potentials of all sorted cells were averaged to calculate the average firing rate. The interwave interval is the time between the two subsequent waves determined manually based on the spike trains.

For light-evoked directional selective responses, self-programmed software was used to calculate the peak spike frequency of leading and trailing edge (ON and OFF) responses for each cell. The responses were averaged from 30 recording trials. The directional selectivity index (DSI) was calculated as:



Targeted deletion of Barhl2 and retinal expression of lacZ knock-in reporter gene

To investigate the role of Barhl2 in retinal development, we generated a targeted deletion of Barhl2 by homologous recombination (Fig. S1A,B). The Barhl2lacZ knock-in allele was created by replacing 88 bp in Exon 1 with the lacZ reporter gene and SV40 polyadenylation sequences to prevent the transcription of the remaining Barhl2 coding sequences. In situ hybridization using Barhl2 anti-sense probe showed that the expression of Barhl2 was abolished in Barhl2lacZ/lacZ (Barhl2-null) retinas and anti-lacZ immunostaining revealed the expression of lacZ in the embryonic retinas (Fig. S1C). The heterozygous Barhl2lacZ/+ mice were indistinguishable from the wild type (WT) and therefore included as controls. Offspring from the Barhl2lacZ/+ intercross were born in a standard Mendelian ratio. Out of the first 289 mice weaned, 22.5% were WT, 52.9% were Barhl2lacZ/+, and 24.6% were Barhl2lacZ/lacZ. Barhl2-nulls (Barhl2lacZ/lacZ) mice were indistinguishable from littermate controls at birth. However they did not gain weight at the same rate as controls and gradually displayed signs of impaired motor coordination. Bahrl2-null mice died prior to postnatal day 24 (P24).

Previous studies have shown that the earliest expression of Barhl2 in mouse retina is at E13.5 in the inner neuroblastic layer (NBL). In the adult retina, Barhl2 is expressed in amacrine, horizontal and ganglion cells (Mo et al., 2004). To examine the retinal cell types that Barhl2-lacZ is expressed in and to determine whether the expression pattern of lacZ reflects that of endogenous Barhl2, we double-immunolabeled retinas from Barhl2lacZ/+ mice with anti-lacZ antibody and specific cell type-specific markers. LacZ+ cells were in three distinct locations in the retina, in the GCL and the inner and outer portions of the INL. In the GCL, LacZ+ cells colabeled with a RGC marker, POU4F1, in some RGCs (Fig. S1D). All Barhl2 expressing cells in the GCL and the inner portions of the INL co-labeled with PAX6, a marker for all RGCs and ACs (Fig. S1E). Double staining with amacrine subtype-specific markers revealed that lacZ was expressed in all cholinergic ACs (ISL1+), selective glycinergic amacrine (GLYT1+), and GABAergic (GAD65+) ACs (Fig. S1FH). At the outer border of the INL, all horizontal cells (calbindin+) expressed lacZ (Fig. S1I). The absence of co-labeling of lacZ with CHX10, opsin, and GFAP excluded the possibility that Barhl2 is expressed in bipolar, photoreceptor and glial cells, respectively (data not shown). These results are consistent with the expression pattern of Barhl2 previously reported (Mo et al., 2004) and indicated that the knock-in lacZ allele recapitulated endogenous Barhl2 expression.

Retinal defects in Barhl2-null mice

To investigate the role of Barhl2 in retinal development, we first assessed the retinal histology of Barhl2-null retinas at different developmental stages. Prior to E15.5, no alterations in retinal laminar organization and thickness were apparent in Barhl2-nulls (Fig. 1A). Starting at E16.5, Barhl2-null retinas had a noticeable reduction in GCL thickness and by P7 the INL thickness was also reduced (Fig. 1BD). While the overall laminar structure of Barhl2-null retinas resembled those of the controls at the completion of retinogenesis (at P21), Barhl2-null retinas had 16.3 ± 2% fewer cells in the GCL and 35.6 ± 6% less cells in the INL (Fig. 1EG).

Figure 1
Developmental abnormality of Barhl2-null retinas. Retinas from control and Barhl2-null mice were analyzed by haematoxylin and eosin (H&E) staining at different developmental stages. Compare to the controls, no overt change in retinal thickness ...

The expression of Barhl2 in amacrine, horizontal and ganglion cells, and the marked loss of cells in the GCL and INL in Barhl2-null retinas suggested that deletion of Barhl2 could impair the differentiation and or survival of neurons in the GCL and INL. To test this possibility, cell type specific markers were used to analyze the development of retinal cell types and subtypes. Immunostaining for POU4F1, which labels approximately 70% of RGCs (Xiang et al., 1993), revealed a 35% decrease of these cells in the GCL (Fig. 2A). In the INL, anti-PAX6 labeling revealed a 39% loss of ACs (Fig. 2B). No significant changes in horizontal (calbindin+), bipolar (CHX10+) and photoreceptor (Opsin+) cells (Fig. 3C–F). In summary, the targeted deletion of Barhl2 selectively affects the development of RGCs and ACs but not horizontal cells that also express Barhl2.

Figure 2
Effect of Barhl2-null mutation on RGCs and ACs in P21 retinas. Retinas were immunolabeled with indicated antibodies (green) and nuclear counter-stained with propidium iodide (red). A, Loss of Barhl2 leads to a severe loss of RGCs immunoreactive for POU4F1 ...
Figure 3
Analysis of RGCs and displaced ACs in the GCL of Barhl2-null retina. AD, Retinas from control and Barhl2-null mice at P21 were co-immunolabeled with anti-PAX6 (red) and anti-POU4F2 (green) (A,B) or with anti-ISL1 (red) and anti-POU4F2 (green) ...

Developmental loss of RGCs in Barhl2-null retinas

The loss of RGCs in Barhl2-null retinas prompted us to analyze further the changes in cell composition in the GCL. Whole mount immunostaining of the adult retinas revealed an approximate 47% decrease of POU4F2+ RGCs (Fig. 3A–E) and an 13.8 ± 4% decrease in PAX6+ cells (Fig. 3A,B,E) in Barhl2-nulls. While ISL1 immunostaining for RGCs and cholinergic ACs revealed a 28.3 ± 3.0% loss of all ISL1+ cells in the GCL, the POU4F2- and ISL1+ cells (presumably the displaced cholinergic ACs) increased by more than two-fold (Fig. 3C–E), suggesting that the loss of RGCs in Barhl2-nulls is accompanied by an increase in the displaced ACs in the GCL, particularly, the cholinergic amacrine subtypes.

To investigate whether the loss of RGCs results from a dysregulation of neurogenesis or cell death, we next tested the generation of RGCs in the absence of Barhl2. POU4F2 expression marks the majority of RGCs during embryogenesis (Young, 1985; Gan et al., 1999). Immunostaining of POU4F2 revealed no change in the number of RGCs in Barhl2-null retinas before E15.5 (Fig. 4A,B,M), suggesting that the genesis of RGCs was not affected in Barhl2-nulls. However, a 19.4% decrease of POU4F2+ RGCs was observed starting at E16.5 in Barhl2-null retinas compared to controls (Fig. 4C,D,M). The number of POU4F2+ RGCs continued to decrease with age (Fig. 4E–H,M). Consistently, the reduction of RGCs was confirmed by a 33.7% reduction in the cross section areas of the Barhl2-null optic nerves (Fig. 4I–L). To determine whether the loss of RGCs resulted from apoptosis in Barhl2-null retinas, we immunolabeled retinas with anti-activated CASP3, one of the main effectors in developmental cell death (Kuida et al., 1996). Barhl2-null retinas showed a gradual increase of apoptotic cells in the GCL from E16.5 to P0 compared to controls (Fig. S2), suggesting that the newly generated RGCs undergo apoptosis in Barhl2-null retinas.

Figure 4
Loss of RGCs in Barhl2-null retina. A and B, No overt change in the number of POU4F2 immunoreactive RGCs is detected in Barhl2-null retinas at and before E15.5. CH, Starting at E16.5, there is a progressive reduction in RGCs in Barhl2-null retinas. ...

Previous studies have shown that BARHL2 is expressed in approximate 33% of RGCs (Mo et al., 2004). The loss of RGCs in Barhl2-nulls is of comparable number, suggesting a cell-autonomous role of BARHL2 in RGCs. Co-immunolabeling Barhl2-null retinas using anti-POU4F2 and anti-lacZ demonstrated that 90–95% of RGCs immunoreactive for lacZ and POU4F2 were absent in Barhl2-null retinas (Fig. 4N–O, arrows). In contrast, the number of RGCs immunoreactive for POU4F2 alone (green) remained the same, consistent with the cell-autonomous role for BARHL2 in development and survival of a selective set of RGCs.

Although BARHL2 expression is restricted to a subset of RGCs in the adult, it may also be expressed transiently in other RGCs during development. To investigate the fate of all Barhl2 lineage RGCs, we employed a lineage tracing strategy to permanently mark cells that ever expressed Barhl2. Barhl2-Cre knock-in mice were generated and expressed Cre recombinase under Barhl2 regulatory sequences. Barhl2-Cre mice were bred with the Z/EG conditional GFP reporter mice, which constitutively expresses an enhanced GFP after the Cre-mediated recombination (Novak et al., 2000), to generate Barhl2cre/+; Z/EG/+ mice for the analysis of Barhl2-lineage cells in control retinas (Fig. S3). Similarly, Barhl2-null (Barhl2cre/lacZ; Z/EG/+) mice were generated from Barhl2cre/+; Z/EG/+ and Barhl2lacZ/+ heterozygotes crossing to determine Barhl2-lineage cells in the absence of Barhl2. Co-labeling of GFP and POU4F2 demonstrated that compared with heterozygous retina at P21, Barhl2-null retinas had a loss of ~95% of Barhl2-lineage RGCs (Fig. 4P,Q), thus confirming the cell-autonomous role of Barhl2 and the necessity of Barhl2 in the development of RGCs.

Altered amacrine subtype composition in Barhl2-null retinas

The partial loss of ACs in the INL (Fig. 2B) and the two-folds increase in POU4F2- and ISL1+ cells (presumably the displaced cholinergic ACs) in the GCL (Fig. 3C–E) prompted us to examine the subtype-specific changes of ACs in Barhl2-null retinas. Immunostaining of GAD65 for GABAergic ACs showed that the number of GABAergic ACs was significantly reduced in Barhl2-null retinas by 34% compared to the controls (Fig. 5A). Immunolabeling of BHLHB5, a bHLH TF expressed in a group of GABAergic ACs (Feng et al., 2006), revealed a similar loss of GABAergic ACs (Fig. 5B). Likewise, anti-GLYT1 labeling demonstrated a 40% reduction in glycinergic ACs in Barhl2-null retinas compared to the controls (Fig. 5C). In contrast, the cholinergic ACs labeled by anti-ISL1, ChAT and calretinin in the ACL were increased (Fig. 5D–F). Whole-mount immunolabeling showed that in the ACL, ISL1 expression completely overlapped that of ChAT and that there was a three-fold increase in the cholinergic amacrine cell density in Barhl2-null retinas (Fig. 6A,B). Similarly, ChAT- and ISL1-labeled displaced cholinergic ACs were increased by two folds (Fig. 6C,D).

Figure 5
Subtype-specific effects on ACs Barhl2-null retina. Sections from P21 mouse retinas were immunolabeled with ACs subtype-specific markers (green) and nuclear counter-stained with PI (red). Loss of Barhl2 leads to a severe loss of ACs immunoreactive for ...
Figure 6
Cholinergic ACs are increased in Barhl2-null retinas. A and B, Double-immunolabeling of control and Barhl2-null retinas with anti-ISL1 (green) and anti-ChAT (red) reveals a three-fold increase in cholinergic ACs in the absence of Barhl2 at P21. C and ...

ISL1 is expressed in developing and mature cholinergic ACs. Retina-specific deletion of Isl1 abolishes all cholinergic ACs (Elshatory et al., 2007). To elucidate the mechanism underlining the increase of cholinergic ACs in Barhl2-null retinas, we assessed ISL1 expression during early postnatal retinal development. We found that there were approximate two-fold increases in the number of cells expressing ISL1 at the inner edge of the NBL (presumably the developing ACL) in Barhl2-null retinas at P0 to P5 (Fig. 6E,F, brackets). In contrast, there was a reduction of 40% differentiating PAX6+ ACs in Barhl2-null retinas at P5 (Fig. 7A,B). Anti-BHLHB5 immunolabeling detected a similar reduction of a subset of GABAergic ACs at this stage (Fig. 7C,D). Moreover, using anti-activated CASP3 immunostaining, we detected an increase of apoptotic cells in the NBL where the nascent ACs were localized (Fig. 7E,F,J). Co-labeling with anti-PAX6 and ati-CASP3 confirmed some of these apoptotic cells as ACs (Fig. 7G–I, arrows). Thus, loss of Barhl2 affects the amacrine subtype composition by differentially changing the genesis and survival of different amacrine subtypes.

Figure 7
Barhl2-null retinas exhibit an elevated cell apoptosis in ACs. A and B, Immunostaining of retina sections with PAX6 reveals a significant loss of PAX6+ amacrine cells (brackets) in Barhl2-null retina at P5. C and D, Anti-BHLHB5 immunostaining shows a ...

Roles of Barhl2 in the genetic regulatory networks of ACs and RGCs

Neurod1 and Neurod4 are redundantly required for AC genesis (Inoue et al., 2002). To define the role of Barhl2 in the genetic regulatory network of AC cell differentiation, we analyzed the expression of Neurod4 and Neurod1 in control and Barhl2-null retinas by in situ hybridization and observed no detectable alteration in their expression from E14.5 to P0 (Fig. 8A,B and data not shown), suggesting that Barhl2 unlikely functions upsteam of Neurod4 and Neurod1 during AC differentiation. Rather, Barhl2 could function downstream of Neurod4 and Neurod1 in the AC differentiation pathway to negatively regulate the development of cholinergic amacrine subtype by suppressing Isl1.

Figure 8
Analysis of potential upstream and downstream genes of Barhl2. AC, The expression profiles of retinogenic bHLH genes, Atoh7, Neurod4 and Neurod1, are unaffected in Barhl2-null retinas by in situ hybridization at E16.5. DE, Barhl2 expression ...

In Drosophila, the proneural gene ato is required for the specification of the first born retinal neurons, R8 photoreceptors (Brunet and Ghysen, 1999). During retinogenesis, ato expression is sequentially restricted from the most anterior region of morphogenetic furrow to R8 founder cells. This restricted expression of ato is partially controlled by BarH1 and BarH2. Loss of BarH induces cell-autonomous ectopic expression of ato (Lim and Choi, 2004). The murine orthologue of ato, Atoh7, is transiently expressed in ganglion-competent retinal precursors and is essential for RGC differentiation (Wang et al., 2001; Yang et al., 2003). Thus, we asked whether Barhl2 could negatively regulate the expression of Atoh7. As shown by in situ hybridization, loss of Barhl2 had no effect on the expression of Atoh7 (Fig. 8C). Conversely, the expression of Barhl2 was significantly down-regulated in both the GCL and the inner NBL of Atoh7-null retinas at E13 (Fig. 8D). Further, Barhl2 expression was greatly down-regulated in Pou4f2-null retinas at E13.5 (Fig. 8E). Therefore, Barhl2 functions downstream of Atoh7-Pou4f2 pathway to regulate the development of a subset of RGCs.

Functional impairments in Barhl2-null mice

Finally, we determined the functional impacts of Barhl2 deletion on retinal cholinergic ACs by recording spontaneous and light evoked synaptic activity of RGCs. We first examined how genetic deletion of Barhl2 affects cholinergic synaptic activity in the developing retina. We used a multielectrode array (MEA) recording to examine spontaneous RGC activity in Barhl2-null mice at P3 and P13 (Fig. 9A). The spontaneous retinal waves at P3 are mediated by nicotinic acetylcholine receptors (nAChRs) and the retinal waves at P13 are mediated by glutamate receptors (GluRs) (Bansal et al., 2000). The frequency of nAChR-mediated retinal waves, measured as the interwave interval, in Barhl2-null mice at P3 was about 50% lower than that of age-matched controls, while the interwave interval of GluR-mediated retinal waves in Barhl2-null mice at P13 was not different from age-matched controls (Fig. 9B–D), demonstrating that genetic deletion of Barhl2 preferentially affects nAChRs-mediated spontaneous synaptic transmission in developing retina.

Figure 9
The retinal waves mediated by cholinergic synaptic transmission were affected in Barhl2-null mice. Spontaneous retinal waves were recorded from P3 and P13 Barhl2-null mice and age-matched WT controls. A, Example of retinal waves recorded from a P3 WT ...

We then examined whether genetic deletion of Barhl2 affects the light evoked synaptic transmission specific to cholinergic ACs by examining the light evoked responses of directional selective (DS) RGCs. DSRGCs exhibit vigorous spiking activity when a visual object moves in their preferred direction across the receptive field but minimum response when the same object sweeps across in the opposite (null) direction (compare Fig. 10A and B) due to spatially offset GABAergic inhibition from cholinergic starburst ACs (Demb, 2007). In mouse retina, most of the ON-OFF responsive RGCs are DSRGCs. We specifically examined the strength of the directional selectivity, as a measure of the specific asymmetric inhibitory synaptic transmission between cholinergic starburst ACs and DSRGCs, and the peak frequency of light evoked responses, as a measure of the strength of overall inhibitory synaptic inputs, of ON-OFF DSRGCs in Barhl2-null mice. Consistent with an increase of GABAergic synaptic inputs from cholinergic starburst ACs, the average peak frequency of light evoked responses of ON-OFF DSRGCs in Barhl2-null mice is significantly lower than that of controls (Fig. 10D and 10E). In contrast, the average peak frequency of light evoked responses of ON RGCs in Barhl2-null mice, which are dominated by non-DSRGCs, is not different from that of controls (Fig. 10C). Interestingly, the peak frequency of light evoked responses of the null direction of ON-OFF DSRGCs is more significantly suppressed in Barhl2-null mice, which resulted in a slight increase of the strength of directional selectivity (Fig. 10F). This is consistent with an asymmetric GABAergic inhibition from cholinergic starburst ACs to ON-OFF DSRGCs with more effective inhibition on the null direction. Taken together, these results strongly support the notion that genetic deletion of Barhl2 specifically impaired GABAergic synaptic transmission mediated by cholinergic starburst ACs in mouse retina.

Figure 10
Light evoked responses of ON-OFF DSRGCs are impaired in Barhl2-null mice. Responses to the moving bars of 12 different directions were recorded from the cells located in the RGC layer of P15 Barhl2-null mice and age-matched WT controls. The cells were ...


In this report, we have generated Barhl2-null mice and demonstrated Barhl2’s essential role in the development of multiple retinal cell types. Loss of Barhl2 results in a marked decrease of RGCs. During the development of ACs, Barhl2-null mutation affects the development of distinct AC subtypes. Loss of Barhl2 leads to the partial loss of glycinergic and GABAergic ACs, a significant increase in cholinergic ACs, and altered synaptic inputs of cholinergic ACs to RGCs in both immature and mature retinas. Gene expression studies show that Barhl2-null mutation leads to the up-regulation of ISL1, the key regulator of cholinergic ACs in the INL and the GCL. Furthermore, we have demonstrated that Barhl2 functions downstream of Atoh7-Pou4f2 pathway in RGC development and likely downstream of Neurod1 and Neurod4 in differentiating ACs. The differential effect of Barhl2-null mutation on each retinal neuronal subtype implies a mechanism that BARHL2 and other TFs form a unique combinatorial code in each subtype and regulate the acquisition of subtype identities.

Essential roles of Barhl2 in the development of RGCs

Previous studies have demonstrated that ATOH7 and POU4F2 constitute the essential genetic cascade of RGC development. ATOH7 determines the RGC competence of retinal precursors (Brown et al., 2001; Wang et al., 2001; Yang et al., 2003) whereas POU4F2 is required for the terminal differentiation of RGCs including axon growth and pathfinding as well as survival (Gan et al., 1996; Gan et al., 1999). The down-regulation of Barhl2 expression in Atoh7-null and Pou4f2-null retinas (Fig. 8D,E) and the partial loss of RGCs in Barhl2-null retinas suggest that Barhl2 functions downstream of Atoh7 and Pou4f2 to control the differentiation of a subset of RGCs. Similarly, in Xenopus, Xbh1 (orthologous to BARHL2) is reported to interact with Xath5 and Xath3 and act as a late transcriptional repressor downstream of Xath5 in the RGCs development (Poggi et al., 2004). While Barhl2 is a direct downstream target of MATH1 in the spinal cord (Saba et al., 2005), it remains to be determined whether ATOH7 directly regulates Barhl2 in retinas. However, the late onset of Barhl2 expression at E13.5 and the down-regulation of Barhl2 in Pou4f2-null retinas suggest that Barhl2 is an indirect downstream target of Atoh7 during RGC development.

Targeted deletion of Barhl2 results in a progressive reduction of approximate 35% RGCs and in the reduced thickness of optic nerves in Barhl2-nulls (Fig. 24). The extent of RGC loss is in proportion to the number of RGCs expressing Barhl2 as shown by the co-localization of POU4F2 and lacZ in Barhl2lacz/+ retinas and by lineage analysis using Barhl2-Cre knock-in allele, indicating a cell-autonomous role of Barhl2 in the differentiation and survival of some RGCs. It is plausible that these RGCs belong to particular ganglion subtype(s). Future morphological, physiological, and functional analyses of Barhl2-expressing cell lineage will shed light on its possible role in determining RGC subtypes.

The roles of Barhl2 in the subtype specification of ACs

ACs form synapses with bipolar cells and RGCs and modulate the signal processing by excitatory and inhibitory effects. The development of ACs depends on several TFs. FOXN4 regulates amacrine genesis by activating the expression of amacrine differentiation factors, Neurod4 and Neurod1 (Li et al., 2004). PTF1, a bHLH TF, is essential for the specification of amacrine and horizontal cells, acting as a primary target of FOXN4. These above TFs do not regulate the subtype specification of ACs and little is known about how the mammalian retina generates the numerous different subtypes of ACs. PAX6, a key regulator in the early retinal development, is expressed in differentiating ACs and positively regulates the formation of glycinergic ACs (Marquardt et al., 2001). Recently, we show that BHLHB5 and ISL1 are uniquely expressed and required for the differentiation of specific AC subtypes, GABAergic (Feng et al., 2006) and cholinergic (Elshatory et al., 2007) ACs, respectively. In this study, we demonstrate that Barhl2 is expressed in all amacrine subtypes. Interestingly, loss of Barhl2 does not affect the initial formation of ACs expressing Neurod1 and Neurod4. Rather, it has subtype-specific effects on ACs, resulting in a 40% loss of glycinergic and GABAergic ACs, and in a significant increase of cholinergic ACs. These differential effects imply that BARHL2’s role in amacrine cell differentiation depends on the unique cellular environment of each amacrine subtypes. Therefore, our data indicate that the subtype-specification of ACs is determined after the acquisition of pan-amacrine properties and is governed by the unique combinatorial activities of overlapping transcriptional activators and repressors. Our results of ISL1’s positive role (Elshatory et al., 2007) and BARHL2’s negative role (Fig. 5 and and6)6) in regulating cholinergic AC differentiation are the first example of a combinatory TF code that specifically affects AC subtype differentiation. Future identification of additional regulators of AC differentiation will increasingly elucidate the precise mechanism by which combinatorial TF codes determine the AC subtypes.

Published studies have shown that forced expression of Barhl2 by viral infection in mouse retinas promotes the differentiation of glycinergic ACs but has no effect on the differentiation of GABAergic ACs, RGCs and horizontal cells (Mo et al., 2004). The reported non-essential role of BARHL2 in horizontal cells is consistent with our targeted disruption study and with our model that retinal cell types are regulated by unique combinatory TF codes. However, the effect of ectopic Barhl2 expression on ACs and RGCs appears to contradict the loss of function data presented here showing that Barhl2-null mutation results in a decrease in GABAergic ACs and an increase in cholinergic ACs. One likely explanation is that the viral infection is performed in neonatal retinas at P0 after the differentiation of most ACs and RGCs and in a different extrinsic retinal environment. Additionally, the ectopic expression of BARHL2 is driven by a strong viral promoter and the likely high level of ectopic BARHL2 may not reconstitute the proper combinatory code similar to the endogenous one, thus resulting in a different outcome in the development of RGCs and ACs.

Supplementary Material


We thank Dr. Alexandra Joyner for the W4 mouse ES cells, Dr. Amy Kiernan, and the members of the Gan Laboratory for many helpful discussions and technical assistance. This work was supported by NIH grants EY013426 to L.G., EY012345 to N.T., and the Research to Prevent Blindness challenge grant to Department of Ophthalmology at University of Rochester.


  • Auerbach W, Dunmore JH, Fairchild-Huntress V, Fang Q, Auerbach AB, Huszar D, Joyner AL. Establishment and chimera analysis of 129/SvEv- and C57BL/6-derived mouse embryonic stem cell lines. Biotechniques. 2000;29:1024–1028. 1030–1032. [PubMed]
  • Bansal A, Singer JH, Hwang BJ, Xu W, Beaudet A, Feller MB. Mice lacking specific nicotinic acetylcholine receptor subunits exhibit dramatically altered spontaneous activity patterns and reveal a limited role for retinal waves in forming ON and OFF circuits in the inner retina. J Neurosci. 2000;20:7672–7681. [PubMed]
  • Brown NL, Patel S, Brzezinski J, Glaser T. Math5 is required for retinal ganglion cell and optic nerve formation. Development. 2001;128:2497–2508. [PMC free article] [PubMed]
  • Brunet JF, Ghysen A. Deconstructing cell determination: proneural genes and neuronal identity. Bioessays. 1999;21:313–318. [PubMed]
  • Demb JB. Cellular mechanisms for direction selectivity in the retina. Neuron. 2007;55:179–186. [PubMed]
  • Elshatory Y, Everhart D, Deng M, Xie X, Barlow RB, Gan L. Islet-1 controls the differentiation of retinal bipolar and cholinergic amacrine cells. J Neurosci. 2007;27:12707–12720. [PMC free article] [PubMed]
  • Feng L, Xie X, Joshi PS, Yang Z, Shibasaki K, Chow RL, Gan L. Requirement for Bhlhb5 in the specification of amacrine and cone bipolar subtypes in mouse retina. Development. 2006;133:4815–4825. [PMC free article] [PubMed]
  • Gan L, Wang SW, Huang Z, Klein WH. POU domain factor Brn-3b is essential for retinal ganglion cell differentiation and survival but not for initial cell fate specification. Dev Biol. 1999;210:469–480. [PubMed]
  • Gan L, Xiang M, Zhou L, Wagner DS, Klein WH, Nathans J. POU domain factor Brn-3b is required for the development of a large set of retinal ganglion cells. Proc Natl Acad Sci U S A. 1996;93:3920–3925. [PMC free article] [PubMed]
  • Higashijima S, Kojima T, Michiue T, Ishimaru S, Emori Y, Saigo K. Dual Bar homeo box genes of Drosophila required in two photoreceptor cells, R1 and R6, and primary pigment cells for normal eye development. Genes Dev. 1992;6:50–60. [PubMed]
  • Inoue T, Hojo M, Bessho Y, Tano Y, Lee JE, Kageyama R. Math3 and NeuroD regulate amacrine cell fate specification in the retina. Development. 2002;129:831–842. [PubMed]
  • Kong JH, Fish DR, Rockhill RL, Masland RH. Diversity of ganglion cells in the mouse retina: unsupervised morphological classification and its limits. J Comp Neurol. 2005;489:293–310. [PubMed]
  • Kuida K, Zheng TS, Na S, Kuan C, Yang D, Karasuyama H, Rakic P, Flavell RA. Decreased apoptosis in the brain and premature lethality in CPP32-deficient mice. Nature. 1996;384:368–372. [PubMed]
  • Li S, Mo Z, Yang X, Price SM, Shen MM, Xiang M. Foxn4 controls the genesis of amacrine and horizontal cells by retinal progenitors. Neuron. 2004;43:795–807. [PubMed]
  • Lim J, Choi KW. Induction and autoregulation of the anti-proneural gene Bar during retinal neurogenesis in Drosophila. Development. 2004;131:5573–5580. [PubMed]
  • Lin B, Wang SW, Masland RH. Retinal ganglion cell type, size, and spacing can be specified independent of homotypic dendritic contacts. Neuron. 2004;43:475–485. [PubMed]
  • Marquardt T, Ashery-Padan R, Andrejewski N, Scardigli R, Guillemot F, Gruss P. Pax6 is required for the multipotent state of retinal progenitor cells. Cell. 2001;105:43–55. [PubMed]
  • Masland RH. The fundamental plan of the retina. Nat Neurosci. 2001a;4:877–886. [PubMed]
  • Masland RH. Neuronal diversity in the retina. Curr Opin Neurobiol. 2001b;11:431–436. [PubMed]
  • Mo Z, Li S, Yang X, Xiang M. Role of the Barhl2 homeobox gene in the specification of glycinergic amacrine cells. Development. 2004;131:1607–1618. [PubMed]
  • Novak A, Guo C, Yang W, Nagy A, Lobe CG. Z/EG, a double reporter mouse line that expresses enhanced green fluorescent protein upon Cre-mediated excision. Genesis. 2000;28:147–155. [PubMed]
  • Pan L, Yang Z, Feng L, Gan L. Functional equivalence of Brn3 POU-domain transcription factors in mouse retinal neurogenesis. Development. 2005;132:703–712. [PubMed]
  • Pan L, Deng M, Xie X, Gan L. ISL1 and BRN3B co-regulate the differentiation of murine retinal ganglion cells. Development. 2008;135:1981–1990. [PMC free article] [PubMed]
  • Patterson KD, Cleaver O, Gerber WV, White FG, Krieg PA. Distinct expression patterns for two Xenopus Bar homeobox genes. Dev Genes Evol. 2000;210:140–144. [PubMed]
  • Poggi L, Vottari T, Barsacchi G, Wittbrodt J, Vignali R. The homeobox gene Xbh1 cooperates with proneural genes to specify ganglion cell fate within the Xenopus neural retina. Development. 2004;131:2305–2315. [PubMed]
  • Saba R, Johnson JE, Saito T. Commissural neuron identity is specified by a homeodomain protein, Mbh1, that is directly downstream of Math1. Development. 2005;132:2147–2155. [PubMed]
  • Tian N, Copenhagen DR. Visual stimulation is required for refinement of ON and OFF pathways in postnatal retina. Neuron. 2003;39:85–96. [PubMed]
  • Torborg CL, Hansen KA, Feller MB. High frequency, synchronized bursting drives eye-specific segregation of retinogeniculate projections. Nat Neurosci. 2005;8:72–78. [PMC free article] [PubMed]
  • Wang SW, Kim BS, Ding K, Wang H, Sun D, Johnson RL, Klein WH, Gan L. Requirement for math5 in the development of retinal ganglion cells. Genes Dev. 2001;15:24–29. [PMC free article] [PubMed]
  • Wong RO, Meister M, Shatz CJ. Transient period of correlated bursting activity during development of the mammalian retina. Neuron. 1993;11:923–938. [PubMed]
  • Xiang M, Zhou L, Peng YW, Eddy RL, Shows TB, Nathans J. Brn-3b: a POU domain gene expressed in a subset of retinal ganglion cells. Neuron. 1993;11:689–701. [PubMed]
  • Yang Z, Ding K, Pan L, Deng M, Gan L. Math5 determines the competence state of retinal ganglion cell progenitors. Dev Biol. 2003;264:240–254. [PubMed]
  • Young RW. Cell differentiation in the retina of the mouse. Anat Rec. 1985;212:199–205. [PubMed]


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...