ZNF804A protein in mouse and human brain
In the human brain, the expression of ZNF804A mRNA increases from the embryonic to the early fetal stage and reaches a peak around the early-mid fetal stage. It decreases afterward and stays at a constant level until late adulthood . As mRNA may not correlate with protein level, we first examined ZNF804A protein distribution in the brain. Consistent with our previous study, we observed the ZFP804A expression in primary cultured mouse neurons (Figure 1A). In the adult mouse brain. ZFP804A is expressed in neurons (Figure 1B) not in astrocytes (Figure 1C). Multiple types of neurons showed ZFP804A signal, including dopaminergic neurons (Figure 1D), GABAergic neurons (Figure 1E), and Purkinje neurons (Figure 1F). The immunostaining results confirmed that ZFP804A is localized in the nucleus, cytoplasm, and neurites of CA1 neurons (Figure 1G). Consistently, ZNF804A is localized in both nucleus and cytoplasma of neural cells in human brain sections (Figure 1H). Intriguingly, ZNF804A is highly enriched in some dendrites in human brain (Figure 1I), suggesting a potential role of ZNF804A in neurite development.
ZNF804A knockdown affects the neurite outgrowth and spine formation
ZNF804A protein is enriched in neurites (Figure 1A, 1G, 1I) and we have identified a group of ZNF804A interacting proteins involved in the neurite outgrowth using yeast-2-hybrid system (Y2H)  including FEZ1 and LGALS1. FEZ1 interacts with microtubules to enhance the extension of neurites [14, 15]. LGALS1 encodes galectin-1, which, when oxidized, loses its lectin activity that normally promotes neurite outgrowth and axonal regeneration [16, 17]. The neuronal migration deficits could be rescued by overexpression of the ZNF804A-interacting gene RPSA , indicating that protein–protein interactions could provide potential therapeutic targets for restoring deficits caused by genetic abnormalities. We hypothesize that these interactions can shape the neurite development. To test this hypothesis, we first confirmed the physical interaction of ZNF804A with 3 proteins, LGALS1, FEZ1 and RPSA, by co-immunoprecipitation (Figure 2A).
We next tested the effect of ZNF804A interacting proteins on Zfp804a-mediated neurite outgrowth. We co-transfected a FLAG tag vector or plasmids expressing LGALS1, FEZ1 and RPSA with a shRNA construct that we have shown the efficacy targeting Zfp804a  into primary mouse cortical neurons (Figure 2B-I) and analyzed the effects on dendritic morphology (Figure 3). Zfp804a shRNA significantly suppressed total number of neurite (P=0.0239) (Figure 3A), total dendritic length (P=0.0325) (Figure 3B), and the number of non-primary neurites (P=0.0027) (Figure 3D), but not the number of primary neurite (Figure 3C). Sholl analysis confirmed the defect of neurite outgrowth caused by Zfp804a knockdown (Figure 3F, 3G, 3E).
Intriguingly, LGALS1, rather than FEZ1, rescued the neurite deficits caused by Zfp804a knockdown (Figure 3). LGALS1 recovered total number of neurite branch (P=0.5843), dendritic length (P=0.9999), and number of non-primary neurite (P=0.408) (Figure 3A-D). The Sholl analysis indicated that LGALS1 increased the number of intersections between neurites and the consecutive circles in the Zfp804a downregulation neurons (Figure 3E, 3H-I). FEZ1 failed to rescue the neurite outgrowth deficits, and showed similar pattern as Zfp804a knockdown (Figure 3A-E, 3J-K). RPSA restored total number of neurites (P=0.9981) and non-primary neurites (P=0.7505), but not the dendritic length (P=0.0001) compared to shControl+FLAG condition (Figure 3A-D). The Sholl analysis showed similar complexity of neurite outgrowth in RPSA overexpression group regardless Zpf804a levels (Figure 3E, 3L-M). Multiple comparisons of knockdown experiments were summarized in Supplemental table S1.
Dendritic spines are essential for receiving inputs from synapses. The density and plasticity of dendritic spines play a fundamental role to regulate neural functions. The morphological development of dendritic spines reflects their maturity . A dendritic filopodia with a long neck and small head are considered to be an immature spine. It becomes shorter and reach to its maturity with a morphology of a mushroom head and a short neck . The spine formation was significantly reduced by Zpf804a shRNA (Figure 4A). Downregulation of Zfp804a decreased the total spine density (P=0.0003) and short spines (P<0.0001) (Figure 4B-D), but not long spine number. LGALS1, FEZ1, or RPSA overexpression at basal level (with control shRNA) did not affect total dendritic spine density (Figure 4A-B). Interestingly, LGALS1, FEZ1, or RPSA ameliorated the defect of total spine density (Figure 4B), whereas LGALS1 and FEZ1 reversed the defect of short spine density comparing to the control group (Figure 4C). These data suggest that ZNF804A interacting proteins play different roles in ZNF804A-mediated dendritic/spine development.
ZNF804A overexpression reduces the neurite outgrowth and spine density
We previously demonstrated that overexpression of ZNF804A enhances translation rate  and duplication of ZNF804A is reported in patients with psychiatric illnesses  . To determine the biological function of a high level of ZNF804A in neurons, we co-transfected ZNF804A construct with plasmids expressing LGALS1, FEZ1, or RPSA into primary cortical neurons. Surprisingly, we observed a much stronger phenotype in neurite outgrowth in ZNF804A overexpressing neurons (Figure 5, 6).
Overexpression of ZNF804A significantly reduced the number of dendritic branches (P<0.0001) and total dendritic length of neurons (P<0.0001) (Figure 6A-B). The number of both primary and non-primary neurites were decreased (Figure 6C-D). The Sholl analysis support a significantly reduced number of neurite branches (Figure 6E-G).
LGALS1 promoted overall neurite outgrowth in the control group, significantly increasing the number of dendritic branches (Figure 6A) and dendritic length (Figure 6B). Specifically, LGALS1 overexpression significantly increased the number of non-primary neurites (Figure 6D), rather than primary neurites (Figure 6C). However, LGALS1 overexpression failed to restore aberrant neurite outgrowth in the ZNF804A overexpressing group (Figure 6A-D). The Sholl analysis confirms the insufficiency of LGALS1 in recovering ZNF804A overexpression-medicated dendritic defects (Figure 6E, 6H-I).
FEZ1 is known to directly bind to DISC1 protein, an important schizophrenia risk, to regulate neurite outgrowth . Notably, we found that FEZ1 increased the dendritic length and the number of branches in ZNF804A overexpression group (Figure 6A-B). In vector-expressing cells, FEZ1 stimulates outgrowth in both primary and non-primary neurites similar as LGALS1 (Figure 6C-D). However, only FEZ1 overexpression reversed defecits of primary and non-primary neurites induced by ZNF804A overexpression (Figure 6C–D). Further Sholl analysis showed that FEZ1 overexpression restored the number of intersections between neurites and the proximal consecutive circles (Figure 6E, 6J-K)). This evidence indicated that FEZ1 prevented neurite outgrowth deficits caused by ZNF804A overexpression.
Interestingly, RPSA overexpression recovered the total number of neurites as well as both primary and non-primary neurites (Figure 6A, C-D). However, RPSA alone showed a trend to reduce the total dendritic length (P=0.695), and cannot rescue the deficits caused by ZNF804A overexpression (P=0.0007) (Figure 6B). The Sholl analysis confirmed our observation that the intersections between neurites and the consecutive circles did not reduce until 80 µm circle (Figure 6E, 6L-M).
We analyzed the spine density and morphology in the primary cultured mouse neurons with ZNF804A overexpression. The dendritic spines density showed significant reduction in the ZNF804A overexpressed mouse neurons (P=0.0328) (Figure 7A-B). Further analyzing spine morphology indicated that the short spines (P=0.0229) decreased significantly (Figure 7C) with an intact long/thin spine density (Figure 7D). Surprisingly, LGALS1, FEZ1, and RPSA could recovered the deficits of reduced spine density caused by ZNF804A overexpression (Figure 7A-B). They showed similar number of short spines with a slightly elevated number of long/thin spine only in the PRSA overexpression group (Figure 7C-D). Multiple comparisons of overexpression experiments were summarized in Supplemental table S2.