FAM19A (TAFA): An Emerging Family of Neurokines with Diverse Functions in the Central and Peripheral Nervous System

Abstract

Cytokines and chemokines have diverse and pleiotropic functions in peripheral tissues and in the brain. Recent studies uncovered a novel family of neuron-derived secretory proteins, or neurokines, distantly related to chemokines. The FAM19A family comprises five ~12–15 kDa secretory proteins (FAM19A1–5), also known as TAFA1–5, that are predominantly detected in the central and peripheral nervous system. FAM19A expression in the central nervous system is dynamically regulated during development and in the postnatal brain. As secreted ligands, FAM19A proteins appear to bind to different classes of cell surface receptors (e.g., GPCRs and neurexins). Functional studies using gain- and loss-of-function mouse models established nonredundant roles for each FAM19A family member in regulating diverse physiological processes ranging from locomotor activity and food intake to learning and memory, anxiety- and depressive-like behaviors, social communication, repetitive behaviors, and somatosensory functions. This review summarizes major advances as well as the limitations and knowledge gaps in understanding the regulation and diverse biological functions of this conserved family of neurokines.

Graphical Abstract

INTRODUCTION

Cytokines and chemokines are small (~8–15 kDa) secretory proteins produced by immune and nonimmune cells.¹ Although widely known to play critical roles in immunity,^1,2 they also act directly in the brain regulating neurodevelopment,^3–6 food intake,^7–9 body temperature,¹⁰ learning and memory,^11,12 social behaviors,¹³ anxiety- or depressive-like behaviors,^14,15 and neuropathic pain.¹⁶ Efforts to identify novel gene families encoding secretory proteins using a bioinformatics approach uncovered the previously uncharacterized FAM19A (also known as TAFA) family of five conserved proteins distantly related to the chemokine macrophage inflammatory protein 1α (MIP-1α/CCL3).¹⁷ All five family members are similar in size (125–135 amino acids), possess a signal peptide at the N-terminus for secretion, and have 8–10 highly conserved cysteines with characteristic spacing (Figure 1A,B). Conventional chemokines are grouped into four major families (C, CC, CXC, CX₃C) based on their patterns of conserved cysteine residues.¹⁸ The cysteine spacing in FAM19A family members is reminiscent of chemokines in that they possess C, CC, and CXC motifs. The first four family members (FAM19A1–4) are more closely related to each other and share the same number and spacing of cysteine residues; the last family member (FAM19A5) is more distantly related to the other members, having only 8 instead of 10 cysteines (Figure 1A–C).¹⁷

Figure 1.

The highly conserved FAM19A (TAFA) family. (A) Schematic of human FAM19A proteins. Signal peptide (SP) sequences are shaded light blue. The known signal peptidase cleavage sites for human FAM19A1, FAM19A4, and FAM19A5 are indicated based on published studies.^22,37,38 The predicted signal peptidase cleavage sites (based on SignalIP³⁹) for human FAM19A2 are FAM19A3 are also indicated. All Cys residues are indicated with a ball-and-stick. FAM19A1–4 share the same number (total of 10) and spacing of Cys residues, whereas FAM19A5 has 8 instead of 10 Cys residues. The Asn-106 residue on mouse FAM19A5 is glycosylated.²¹ The % amino acid identity between orthologous mouse (Mus musculus), human (Homo sapiens), chicken (Gallus gallus), frog (Xenopus tropicalis or Xenopus laevis), or zebrafish (Danio rerio) proteins is indicated. Percent identity is based on Clustal Omega alignment⁹⁷ of mature FAM19A proteins (excluding the signal peptide). GenBank accession numbers of proteins used in pairwise comparison are the following: mouse FAM19A1 (NP_877960), human FAM19A1 (NP_998774), chicken FAM19A1 (XP_004944733), frog Fam19a1 (XP_012817436), zebrafish Fam19a1 (XP_009304221); mouse FAM19A2 (NP_001239316), human FAM19A2 (NP_848634), chicken FAM19A2 (XP_001234989), frog Fam19a2 (XP_018111408), zebrafish Fam19a2 (XP_002662575); mouse FAM19A3 (NP_899047), human FAM19A3 (NP_877436), chicken FAM19A3 (015154512), frog Fam19a3 (XP_031752674), zebrafish Fam19a3 (XP_005167029); mouse FAM19A4 (NP_796207), human FAM19A4 (NP_872328), chicken FAM19A4 (XP_004944736), frog Fam19a4 (XP_002940559), zebrafish Fam19a4 (NP_001070741); mouse FAM19A5 (NP_001239239), human FAM19A5 (NP_001076436), chicken FAM19A5 (XP_025009572), frog Fam19a5 (XP_012813881), and zebrafish Fam19a5 (XP_005168594). (B) Alignment of mature human FAM19A1–5 (excluding the signal peptide) using Clustal Omega.⁹⁷ All cysteine residues are indicated with a ball-and-stick. Identical and similar amino acids are shaded black and gray, respectively. (C) Amino acid identity between different human FAM19A paralogs, calculated based on NCBI pairwise sequence alignment tool.⁹⁸

FAM19A proteins are highly conserved throughout vertebrate evolution, with a striking degree of amino acid identity among orthologs from divergent species (Figure 1A). This remarkable conservation is consistent with the important central nervous system (CNS) function ascribed to this group of proteins (discussed below). Conservation between different FAM19A family members, however, is lower, supporting the idea that each family member evolved to play unique and nonredundant roles in the CNS (discussed below). Given the predominant expression and diverse functions within the central and peripheral nervous system, this group of neuron-derived secretory proteins is referred to as neurokines.

PREDOMINANT AND DYNAMIC EXPRESSION OF FAM19A IN THE CNS

Detailed analysis of transcript expression based on RNA sequencing,^19,20 quantitative real-time PCR,^17,21–24 LacZ reporter knock-in mice,^25,26 and publicly available transcriptomic databases²⁷ collectively indicate the brain is the predominant organ expressing FAM19A1–5 in humans, mice, and rats. Expression of FAM19A1–4 transcripts in peripheral tissues is either negligible or markedly lower than in the CNS.

In mouse and human brain, FAM19A expression is dynamically regulated during development.^22,25,26,28 For example, the Human Brain Transcriptome database²⁸ indicates dynamic expression patterns of FAM19A1 and FAM19A2 in different brain regions from early postnatal periods to adulthood, while expression of human FAM19A3–5 remains stable. Likewise, studies of LacZ reporter knock-in mice show that Fam19a1 and Fam19a5 expression are spatially and temporally regulated during development.^25,26 Single-cell RNA sequencing of the mouse brain reveals that Fam19a1–4 are primarily expressed by neurons in the central and peripheral nervous system, whereas Fam19a5 is expressed by both neuronal and non-neuronal cell populations in the brain (Figure 2A,B).²⁹ Consistent with RNA sequencing data, Fam19a1–4 transcripts are mainly expressed by cultured primary cortical neurons but not astrocytes, whereas Fam19a5 is expressed by both cultured neurons and astrocytes.²¹ Single-cell RNA sequencing has illuminated the expression patterns of Fam19a genes at remarkable resolution.²⁹ For example, although Fam19a is widely expressed in the CNS, some anatomical regions of the brain express only one Fam19a gene, while others express various combinations of two or more (Figure 2A). Similar patterns can also be observed for cell populations based on functional cell groupings (Figure 2B) or single cell clusters (Figure 2C). While all Fam19a transcripts are expressed by neurons, Fam19a5 is the only one that can also be found in proliferating and nonproliferating glia populations in the enteric nervous system (Figure 2B). All Fam19a family members, except for Fam19a3, can be found in excitatory neurons. In contrast, inhibitory neurons express only Fam19a1, Fam19a2, and Fam19a5.²⁹ Transcript levels, however, do not always correspond to protein abundance.³⁰ While most of the expression data are confined to transcript levels, limited proteomics data indicate that the brain has the highest protein abundance of FAM19A1, FAM19A2, and FAM19A5 in both mice and humans.³¹

Figure 2.

Expression of Fam19a in the mouse nervous system. Data for this figure were based on single-cell RNA sequencing of the mouse nervous system²⁹ and can be accessed via mousebrain.org. Anatomical location of cells expressing the Fam19a genes and functional cell type classifications were used to group data for visualization purposes based on the same schemes as published by the Linnarsson lab.²⁹ (A) Anatomical regions of the nervous system that house cells that express or coexpress Fam19a genes.²⁹ (B) Functional cell types that express or coexpress Fam19a genes.²⁹ (C) Data showing distinct single-cell populations clustered by Fam19a expression and coexpression. There are a total of 160 transcriptionally distinct single-cell populations expressing at least one member of the Fam19a family.²⁹ Numbers within the circle represent the total number of single-cell populations that express the designated Fam19a family member(s), as indicated by the text outside the circle.

REGULATION OF Fam19a EXPRESSION

Several physiologic and pathophysiologic changes can induce or suppress Fam19a expression in the brain. Changes in metabolic state induced by fasting and refeeding alter the expression of Fam19a in different brain regions.²¹ For example, Fam19a1 and Fam19a2 expression is markedly upregulated in the cerebellum and hippocampus and downregulated in the cortex of fasted–refed mice relative to overnight fasted animals. In contrast, Fam19a4 expression in the hypothalamus is suppressed in refed mice relative to fasted animals.²¹ Refeeding increases Fam19a5 expression in the cerebellum yet decreases expression in the cortex and hypothalamus relative to the fasted state.²¹ Unlike other family members, Fam19a3 brain expression is not affected by fasting and refeeding; instead, its expression is influenced by circadian rhythms.³² In the hypophysial par tuberalis (involved in seasonal physiology), Fam19a3 transcript level is low at mid-day and high at midnight. This oscillation in Fam19a3 expression is lost in mice lacking the melatonin receptor subtype 1 (MT1).³² Additionally, hypoxia caused by focal cerebral ischemia strikingly upregulates Fam19a3 in the infarcted mouse brain.²³ In the case of Fam19a5, its expression is upregulated by brain injury and inflammatory stimuli (e.g., TNF-α) in the corpus callosum and hypothalamus, respectively.^26,33 Conversely, dehydration and chronic stress downregulate Fam19a5 expression in the hypothalamus and hippocampus, respectively.^24,34 Recent efforts using single-cell RNA sequencing also showed that neuronal expression of Fam19a1 in the mouse visual cortex is regulated by environmental stimuli (e.g., light).³⁵ Further, neuronal activation in hippocampal cultures (induced by GABA_A receptor antagonist) reduces Fam19a1 and Fam19a2 expression.³⁶ Collectively, these observations indicate that Fam19a expression in different regions of the brain is dynamically regulated by internal and external stimuli.

SECRETION AND GLYCOSYLATION OF FAM19A PROTEINS

All human and mouse FAM19A proteins have an N-terminal signal peptide and are secreted into conditioned medium in vitro when expressed in heterologous cells.^{17,21–23,37} N-terminal sequencing of mature proteins indicates that the signal peptide in FAM19A1,²² FAM19A4,³⁷ and FAM19A5³⁸ is longer than predicted by the commonly used SignalIP program.³⁹ Mature secreted FAM19A1–4 proteins are ~12 kDa in size, whereas secreted FAM19A5 is larger (~15 kDa) due to the presence of N-linked glycans on Asn-106.²¹ In contrast to previous reports,^21,22 recent works by Khalaj and colleagues³⁶ show that FAM19A1 is poorly secreted when expressed in heterologous HEK 293T cells. However, secretion of FAM19A1 from cells can be markedly enhanced in the presence of neurexins, as the two proteins form stable complexes via intermolecular disulfide bonds during biosynthesis and transport through the secretory pathway.³⁶ Neurexins are important presynaptic adhesion molecules on the plasma membranes of neurons that bind diverse trans-synaptic ligands.⁴⁰ Deletion of all three endogenous neurexin isoforms (NRXN1–3) results in a significant reduction, but not a complete block, of surface-exposed FAM19A1 in mouse hippocampal cultures;³⁶ this observation suggests that neurexins promote, but are not required for, FAM19A1 transport to the extracellular surface of neurons. Unlike the rest of the family, FAM19A5 does not associate with neurexins.³⁶

BIOLOGICAL FUNCTIONS OF FAM19A

Since its initial discovery,¹⁷ the functions of the FAM19A family of brain-enriched neurokines have remained enigmatic. Functional studies of FAM19A fall into two broad categories: assessing the biological effects of recombinant FAM19A treatments in cultured primary cells or cell lines (Table 1) or in mice (Table 2) or determining the functional deficits due to Fam19a inactivation in knockout mouse or zebrafish models (Table 2). Some, but not all, in vitro studies have in vivo correlates. Although the in vitro studies of FAM19A are revealing and suggestive, additional studies are needed to put some of these observations in greater physiological context.

Table 1.

Summary of FAM19A’s Functions Based on in Vitro Cell Model Systems^a

protein	source	purity	epitope tag	effects on cell model system in vitro	ref
FAM19A1	HEK 293Tb	>98%	Myc-His	Promotes neuronal cell differentiation	22
	Lentivirus		V5	Reduces posttranslational modifications of neurexins in hippocampal cultures	36
FAM19A2	E. coli c	>95%	?	Enhances human mesenchymal stem cell proliferation and migration	43
	HEK 293Ed	>95%	None	Inhibits paraventricular nucleus CRH neurons	50
FAM19A3	HEK 293Tb	>98%	None	Promotes microglial polarization and suppresses inflammation	23
FAM19A4	Eukaryotic cellse	?	Myc-His	Enhances macrophage chemotaxis and phagocytosis	37
FAM19A5	E. coli f	>95%	His	Promotes macrophage chemotaxis and inhibits osteoclastogenesis	48
	E. coli	>98%	?	Inhibites vascular smooth muscle cell proliferation and migration	38

^aAll epitope tags are located in the C-terminus of FAM19A proteins, except FAM19A5 made in bacteria has an N-terminal His-tag.

^bSecreted proteins were purified from the conditioned media.

^cFAM19A2 produced by R&D systems.

^dFAM19A2 purified from cell lysate of transfected cells.

^eFAM19A4 was produced by Crown Bioscience. No information on eukaryotic cells used.

^fFAM19A5 produced by Biovendor R&D.

Table 2.

Summary of FAM19A’s Functions Based on Gain- and Loss-of-Function Animal Models^a

gene/protein	model	phenotype	ref
FAM19A1	KO mice	Locomotor hyperactivity, reduced anxiety, altered food intake patterns, reduced sensitivity to electric foot shock, long-term memory deficit	21
		Locomotor hyperactivity, reduced body weight, and long-term memory deficits	25
FAM19A2	WT mice	Central delivery of FAM19A2b increases meal frequency, food intake, and energy expenditure.	55
	KO mice	Elevated anxiety behaviors Impaired spatial and short- and long-term memory Increased anxiety-like behaviors and decreased depression-like behavior Reduced CREB target gene (BDNF, c-Fos, NF1, CBP) expression	50
	KO zebrafish	Increased anxiety-like and fear-like behaviors	50
FAM19A3	WT mice	Intraperitoneal injection of FAM19A3c attenuates cerebral ischemia and neuronal cell death FAM19A3c promotes microglial polarization toward the anti-inflammatory M2 type in brain ischemia model	23
	KO mice	Impaired response to social novelty, deficit in social communication, increased repetitive behaviors, and elevated anxiety	61
FAM19A4	WT mice	Intratechal delivery of FAM19A4(TAFA4)d reversed carrageenan-induced mechanical hypersensitivity	51
		Intratechal injection of FAM19A4d alleviates mechanical pain induced by noxious stimuli in complete Freud adjuvant (CFA)-inflamed paws	63
	KO mice	Enhanced mechanical and chemical hypersensitivity following inflammation and nerve injury	51
FAM19A5	WT mice	Adenoviral-mediated overexpression of FAM19A5 inhibits wire injury-induced neointima formation in femoral arteries	38
		Knockdown of FAM19A5 in the brain attenuates TNF-α induced anorexia, weight loss, and inflammatory gene expression	33
		Central delivery of FAM19A5e induces anorexia, weight loss, hyperthermia and increases inflammatory gene expression	33
		Water deprivation downregulates FRAM19A5 (TAFA5) mRNA in the paraventricular nucleus (PVN); regulates fluid homeostasis?	34
		AAV-mediated overexpression of FAM19A5 in hippocampus attenuates stress-induced depressive-like behaviors	24
	WT rat	Adenoviral-mediated overexpression of FAM19A5 inhibits balloon injury-induced neointima formation in carotid arteries	38
	LacZ knockin mice	Traumatic brain injury upregulates FAM19A5 reporter in a subset of neurons and oligodendrocyte precursor cells (OPCs)	26
	KO mice	Elevated depressive-like behaviors and impaired spatial memory Reduced AMPA and NMDA receptors expression in hippocampus Decreased dendritic spine density in hippocampal neurons	24

^aWT, wild-type; KO, knockout.

^bMade in bacteria (Millipore Sigma; SRP3167); ≥98% purity.

^cMade in mammalian HEK 293T cells and purified from the conditioned medium; >98% purity.

^dMade in bacteria (R&D Systems; 5099-TA); >95% purity.

^eObtained from Neuracle Science (Seoul, South Korea).

In Vitro Studies.

(a). FAM19A1.

Neural stem cells give rise to neurons and glia.⁴¹ Treatment of mouse embryonic brain-derived neurospheres in culture with recombinant FAM19A1 reduces neural stem cell proliferation and self-renewal and promotes differentiation of neural stem cells into neurons while suppressing astrocyte differentiation.²² FAM19A1 appears to regulate neural stem cell proliferation and differentiation by binding to the N-terminal domain of G-protein-coupled receptor 1 (GPR1) and engaging the ROCK/ERK1 pathway to suppress cell proliferation and the ROCK/STAT3 pathway to promote neuronal differentiation.²² This study suggests FAM19A1 to be a novel secreted factor regulating neural stem cell fate. FAM19A1 deficiency was predicted to result in altered neuronal and astrocyte populations in the brain. Targeted inactivation of Fam19a1 in mice, however, does not alter overall size of the brain or affect neuronal or glial cell populations or morphology in cortical layers.²⁵ Subtle changes in cell composition and morphology in other brain region remain to be seen.

FAM19A proteins are pan-neurexin ligands.³⁶ In mouse hippocampal cultures, lentiviral-mediated overexpression of FAM19A1 reduces miniature inhibitory postsynaptic currents. Altered inhibitory synaptic transmission in hippocampal cultures is attributed to FAM19A1-mediated reduction in O-glycosylation and heparan sulfate modifications of neurexins.³⁶ Since interactions occur in the ER/Golgi, it has been suggested that FAM19A1 binding to neurexins in the secretory pathway sterically hinders glycosylation. Given that the stoichiometry of the FAM19A1–neurexin complex is one-to-one, this model would predict that the relative abundance of FAM19A1 might influence the extent of posttranslational modifications of neurexins. This study has highlighted an interesting and a novel mode of regulating protein modifications in neuronal cell adhesion molecules that are likely to impact their functions.

(b). FAM19A2.

Skeletal (stromal or mesenchymal) stem cells play important roles in bone remodeling and recovery from bone fracture.⁴² FAM19A2 transcript and protein levels are upregulated during fracture healing in mice.⁴³ In cultured human skeletal (stromal or mesenchymal) stem cell lines, recombinant FAM19A2 treatment promotes trans-well migration and motility and enhances lamellipodia formation by activating the Rac1-p38 pathway.⁴³ Because local and systemic circulating levels of FAM19A2 are not known, it is unclear whether the dose used (10 μg/mL) in most functional in vitro assays with cultured cells is within the concentration range encountered by cells in an in vivo milieu. The physiologic relevance of the in vitro findings can be evaluated in future studies using Fam19a2 KO mice subjected to bone fracture analysis. Although preliminary, this study suggests a potential role for FAM19A2 in the skeletal system.

(c). FAM19A3.

Microglial cells are key mediators of infection- or trauma-related immune and inflammatory responses in the brain.⁴⁴ In a mouse model of transient focal ischemia, FAM19A3 expression is markedly upregulated in the brain and in microglial cells.²³ Microglial cells can be polarized toward the proinflammatory M1 type or the anti-inflammatory M2 type.⁴⁵ Treatment of primary mouse microglial cells with recombinant FAM19A3 promotes M2 polarization in culture and suppresses inflammatory cytokine (IL-6 and TNF-α) secretion induced by LPS and INF-γ.²³ Consistent with in vitro findings, peripheral delivery of FAM19A3 attenuates brain infarct size induced by transient focal ischemia in mice.²³ Due to its small size, recombinant FAM19A3 delivered via intraperitoneal injection presumably can cross the blood–brain barrier (via either passive or active transport) to exert its effects in the brain, although this has not been demonstrated. This study suggests FAM19A3 to be an injury-responsive factor that acts on microglial cells to dampen inflammatory response in the brain.

(d). FAM19A4.

Macrophages play critical roles in immunity and tissue repair and can become polarized toward the M1 or M2 type.⁴⁶ LPS can markedly induce expression and secretion of FAM19A4 in a macrophage cell line and in primary human monocytes and macrophages.³⁷ In primary human macrophages and resident mouse peritoneal macrophages, recombinant FAM19A4 stimulation promotes phagocytosis and enhances reactive oxygen species (ROS) production in the presence of zymosan.³⁷ FAM19A4 appears to exert its effects on macrophages through formyl peptide receptor 1 (FPR1).³⁷ While FAM19A4 is most prominently expressed in the nervous system in the basal state, this study suggests that the transcript can be induced in immune cells and may have non-neuronal functions in peripheral tissues. This study also suggests macrophages to be a potential cell target of FAM19A4.

(e). FAM19A5.

Bone-marrow derived macrophages (BMDMs) can be induced to differentiate into osteoclasts, which play important roles in bone resorption.⁴⁷ In cultured BMDMs, recombinant FAM19A5 stimulation promotes cell migration and antagonizes RANKL-induced osteoclast differentiation by suppressing expression of osteoclast-related genes and enhancing expression of negative regulators of osteoclastogenesis (e.g., MafB and IRF8).⁴⁸ FAM19A5 appears to act through the formyl peptide receptor 2 (FPR2); its effects on BMDMs are significantly blocked by FPR2 antagonist or Fpr2 inactivation.⁴⁸ This study suggests FAM19A5 to be a novel factor regulating macrophage cell fate and may have a role in regulating the extent of bone turnover in the skeletal system. The in vitro findings can be further confirmed using Fam19a5 KO mice.

In Vivo Studies.

Although FAM19A expression can be detected in the brain during early development,^21,25,26 single gene knockout (KO) mouse models show that none of the FAM19A family members are required for brain development.^{21,24,25,49–51} All KO mice are born at the expected Mendelian ratio, and postnatal mice do not exhibit gross developmental phenotypes. The physiological functions of different FAM19A family members can be gleaned from the salient phenotypes and behavioral deficits in Fam19a gene knockout mouse or zebrafish models. Additional biological insights have been gained from FAM19A overexpression or recombinant FAM19A delivery into wild-type (WT) mice or rats in the context of physiology and pathophysiology, as described below.

(a). FAM19A1.

Targeted inactivation of Fam19a1 results in sexually dimorphic phenotypes in mice.²¹ Although both male and female Fam19a1 KO mice are hyperactive, the level of locomotor hyperactivity is more pronounced in female KO mice. Hyperactivity is attributed to altered dopamine turnover in the striatum of KO animals.²¹ Despite similar caloric intake, Fam19a1 KO male but not female mice have altered food intake behaviors and patterns. Male KO animals eat meals more frequently, consume more food in the light cycle, and ingest less food during the dark cycle compared to WT mice. Due to elevated physical activity levels, Fam19a1 KO female mice have a higher metabolic rate and energy expenditure.²¹ Intriguingly, despite increased energy expenditure, no body weight differences are observed between WT and Fam19a1 KO female mice over time. Thus, FAM19A1 deficiency affects food intake patterns and behaviors, physical activity level, and whole-body energy expenditure in a sex-dependent manner.

Fam19a1 KO mice also display reduced anxiety-like behaviors in open field and elevated plus maze tests.²¹ Short-term spatial working memory (assessed by Y-maze spontaneous alternation tests) and sensorimotor gating (assessed by prepulse inhibition of acoustic startle response) are normal in Fam19a1 KO mice. A trace fear conditioning test, however, reveals a deficit in hippocampal-dependent learning and memory. Fam19a1 is expressed in sensory neurons located in the mouse lumbar dorsal root ganglion (DRG),⁵² hinting at a possible somatosensory function. In support of this, Fam19a1 KO female, but not male, mice appear to have reduced sensitivity to pain induced by electric foot shock or hot plate.²¹

In a separate study, Yong and co-workers²⁵ noted similar hyperactivity and reduced anxiety-like behaviors in Fam19a1 KO mice as well as a deficit in hippocampal-dependent learning and memory. Although Fam19a1 KO mice do freeze less than WT mice in response to conditioned tone tests,^21,25 Lei et al.²¹ observed this phenotype in female but not male KO mice, while Yong et al. observed this phenotype in male KO mice but did not assess female mice.²⁵ The reason for these sex-dependent response discrepancies is unclear and could be the result of differences in experimental settings and protocols. Further, interpretation of the results is partially confounded by locomotor hyperactivity of mice in both studies. There is, however, one major difference between the two studies: in Yong et al.,²⁵ Fam19a1 KO mice had lower body weight relative to WT controls, a phenotype not seen by Lei et al.²¹ Although the mechanism of action of FAM19A1 in the brain remains largely unknown, these two studies using genetic KO mouse models clearly indicate a role for this neurokine in modulating locomotor activity, anxiety-like behaviors, learning and memory, and somatosensory functions.

(b). FAM19A2.

FAM19A2 is distantly related to MIP-1α/CCL3,¹⁷ a chemokine previously reported to act in the hypothalamus to promote fever and suppress food intake.⁵³ A genome-wide association study also identified FAM19A2 as a novel insulin sensitivity locus.⁵⁴ To test whether FAM19A2 acts in the hypothalamus to influence metabolism, Okada and colleagues⁵⁵ delivered recombinant FAM19A2 into the third ventricle of mice just before dark cycle initiation and observed a significant increase in food intake and meal number. They also observed increased energy expenditure and preferential carbohydrate oxidation over fat (judged by increased respiratory exchange ratio).⁵⁵ Central delivery of FAM19A2, however, does not affect physical activity level. These data imply a central role for FAM19A2 in regulating food intake behaviors and energy metabolism, a function shared by FAM19A1.²¹ In zebrafish, fam19a2 (sam2) is exclusively expressed in the CNS, with prominent expression in the dorsal habenula, telencephalon, and hypothalamus.⁵⁰ Inactivation of fam19a2 in zebrafish results in increased anxiety-related behaviors, indicated by increased thigmotaxis (avoidance of open areas), scototaxis (preference for dark over light areas), and shoaling behavior.⁵⁰ Elevation of anxiety-like behaviors in fam19a2 KO zebrafish is attributed to increased expression of corticotropin releasing hormone b (crhb) in the anterior preoptic area, a region corresponding to the hypothalamic paraventricular neurons (PVN) in mammals. Increased anxiety-related behaviors are also observed in Fam19a2 KO mice, indicated by less time spent in open arms of an elevated plus maze as well as increased freezing in contextual and cued fear-conditioning tests.⁵⁰ Mechanistically, FAM19A2 is believed to act in the hypothalamic PVN to regulate CRH neuron excitability. Bath application of recombinant FAM19A2 significantly increases the frequency of spontaneous inhibitory postsynaptic currents in hypothalamic PVN neurons. These results suggest that FAM19A2 modulates tonic GABAergic inputs onto CRH neurons.⁵⁰ The pathophysiological relevance of the findings in KO zebrafish and mice has been highlighted by fine mapping of humans with chromosome 12q14.1 deletion or rearrangement, wherein FAM19A2 is a candidate gene that may contribute to intellectual disability and autism spectrum disorder.^50,56

Using an independent line of Fam19a2 KO mice, Wang et al.⁴⁹ confirmed that Fam19a2 deficiency increased anxiety-like behaviors, assessed by open-field and elevated plus maze tests, a phenotype also observed in an independent KO mouse line generated by Choi et al.⁵⁰ Fam19a2 KO male mice also have reduced depression-like behaviors in forced-swim and tail-suspension tests.⁴⁹ Further, Fam19a2 KO mice have impaired spatial learning and memory indicated by Morris water maze tests and impaired short- and long-term memory determined by novel object recognition tests.⁴⁹ The observed behavioral deficits are attributed to increased neuronal apoptosis in the cortex and hippocampus of the KO mice. Neuronal loss is due, at least in part, to decreased PI3K/AKT and MAPK/ERK signaling and reduced expression of CREB target genes such as BDNF, C-FOS, and NF11.⁴⁹ CREB binding protein (CBP) transcript and protein levels are also reduced in the Fam19a2 KO mouse brain.⁴⁹ These data suggest that FAM19A2 may function as a neurotrophic factor critical for maintaining neuron survival. One limitation of the Wang et al. study is that only male mice were examined.⁴⁹ Because sex affects mouse behaviors,^57–60 as seen in Fam19a1 KO studies,^21,25 it is important to examine whether female Fam19a2 KO mice exhibit similar, no, or exaggerated behavioral phenotypes compared to male KO animals. Choi et al.⁵⁰ do not indicate the sex of KO mice, making it difficult to compare this study’s behavioral phenotypes with those reported by Wang et al.⁴⁹ Additionally, Choi et al. only conducted anxiety-like behavioral tests; neither learning/memory nor depression-like behaviors were assessed in Fam19a2 KO mice,⁵⁰ further precluding side-by-side comparison between the two independent studies.

(c). FAM19A3.

Male, but not female, mice lacking FAM19A3 (SAM3) have increased anxiety, indicated by less time spent in open arms of elevated plus maze tests.⁶¹ Hippocampal-dependent learning and memory are normal in Fam19a3 KO mice of either sex, as determined by trace fear conditioning tests.⁶¹ Interestingly, Fam19a3 KO mice exhibit three major behavioral deficits typically seen in autism spectrum disorders: reduced response to social novelty, impaired social communication, and increased repetitive behavior. In the three-chamber test, Fam19a3 KO mice of either sex spend significantly more time in a chamber with a stranger mouse than in a chamber with a nonsocial novel object, indicating normal sociability.⁶¹ However, when subjected to a social novelty test, Fam19a3 KO mice of either sex show no preference in visiting a stranger mouse over a familiar mouse; in contrast, WT controls or Fam19a3 heterozygous (+/−) mice spend more time with a novel stranger than with a familiar mouse.⁶¹ These results indicate impaired social novelty in mice lacking FAM19A3. Importantly, deficit in social novelty is not due to failure to recognize novel objects. Fam19a3 KO mice appear to have normal empathic responses in the observational fear learning test. However, Fam19a3 KO mice show impaired social communication in a scent marking test.⁶¹ Scent marking in male mice in response to the presence of female urine is considered a form of social communication.⁶² In Fam19a3 KO male mice, the number of scent markings elicited by female urine is greatly reduced compared to WT controls. Lastly, Fam19a3 KO mice exhibit increased repetitive behaviors (e.g., self-grooming and marble burying).⁶¹ These results indicate an association between loss of FAM19A3 and autism spectrum disorder-like behaviors, although the underlying neural mechanisms are currently unknown.

(d). FAM19A4.

FAM19A4 is mostly expressed in sensory neurons in the peripheral nervous system (Figure 2)²⁹ and is a specific marker of C-low-threshold mechanoreceptors (CLTMRs), a specialized subpopulation of cutaneous afferents that mediate touch sensation and whose cell bodies cluster in the DRG and trigeminal ganglia.⁵¹ Applying patch clamp recordings and calcium imaging on genetically marked FAM19A4⁺ DRG neurons (from GFP reporter knock-in mice), Delfini et al.⁵¹ showed that FAM19A4⁺ neurons have properties of C-unmyelinated mechano-nociceptors. Although FAM19A4 is a specific marker for C-LTMRs, its deficiency does not affect C-LTMR developmental specification and function.⁵¹ A battery of tests (rotorod, open field, hot plate, thermotaxis gradient assay, Hargreaves’, cold plate, two temperatures’ choice, dynamic cold- and hot-plate tests) also reveal no abnormalities in motor activity, anxiety, or thermosensation in Fam19a4 KO mice. However, loss of FAM19A4 enhances mechanical and chemical pain hypersensitivity induced by inflammation and nerve injury and increases excitability of spinal cord lamina IIi neurons determined by whole-cell recordings of mouse spinal cord slices; both phenotypes can be reversed by intrathecal delivery or bath application of recombinant FAM19A4.⁵¹ Intrathecal injection of recombinant FAM19A4 also reverses carrageenan-induced mechanical pain hypersensitivity in WT mice.⁵¹ Thus, C-LTMR-derived FAM19A4 may play an analgesic role by modulating the threshold of somatic sensation by controlling neuronal excitability.

A follow-up study showed that FAM19A4 could reinforce inhibitory synaptic transmission within the neuronetwork of the spinal cord of mice.⁶³ Bath application of FAM19A4 activates GABAergic transmission and suppresses local excitatory synapses, dampening synaptic transmission from high-threshold C-fibers. In an inflammatory pain mouse model, the study showed that intrathecal injection of FAM19A4 alleviates mechanical allodynia. The analgesic effects of FAM19A4 could be blocked by antagonizing GABAergic transmission.⁶³ The antinociceptive action of FAM19A4 is partly due to its effects in promoting microglial retraction and increasing the number of inhibitory synapses on lamina IIi neurons.⁶³ Together, these studies^51,63 established a model whereby C-LTMR-derived FAM19A4 acts on GABAergic interneurons and microglia within the spinal cord to modulate nociceptive transmission in pathological contexts. Notably, pain sensitivity is influenced by sex.^64–67 Because both studies used only male mice for all behavioral tests,^51,63 it would be valuable to know if the antinociceptive effects of FAM19A4 in vivo hold true in female mice. The discovery that FAM19A4 is an endogenous secretory protein with analgesic properties raises the therapeutic possibility of using it as a biologic for alleviating chronic inflammatory pain refractory to standard treatment modalities. As a potential pharmacologic agent, it may be worth exploring whether repeated administration of recombinant FAM19A4 to animals results in desensitization to its analgesic effects. Although the somatosensory function of FAM19A4 has yet to be explored in humans, it is interesting to note that methylation of the FAM19A4 promoter is robustly and highly correlated with cervical cancer.^68–73

(e). FAM19A5.

Wang et al.³⁸ identified FAM19A5 as a novel adipokine with markedly higher expression in adipose tissue than in the brain. Expression of Fam19a5 transcript and protein in the visceral (epididymal) fat depot is reduced in genetic and dietary mouse models of obesity and diabetes.³⁸ Because obesity is known to affect vascular health,⁷⁴ Wang and co-workers sought to address whether adipose tissue-derived FAM19A5 acts on the vasculature. Functional studies in vitro revealed that FAM19A5 suppresses vascular smooth muscle cell (VSMC) proliferation and migration.³⁸ Further, adenoviral-mediated overexpression of FAM19A5 in rat carotid arteries or mouse femoral arteries inhibits neointima formation following vascular injury.³⁸ Overexpressing FAM19A5 in adipose tissue of transgenic mice also significantly attenuates neointima formation following wire injury to the femoral arteries.³⁸ Together, these results suggest FAM19A5 to be a secretory factor linking adipose tissue health (e.g., obesity) and the capacity of the vasculature to deal with injury and inflammation. However, the results obtained from transgenic mice should be interpreted with caution since the Fam19a5 transgene is driven by the aP2/FABP4 gene promoter. It has been found that aP2/FABP4 is also expressed in nonadipose tissues (e.g., monocytes, macrophages, brain),^75–78 which may confound interpretation of the results. By a candidate gene approach, sphingosine-1-phosphate receptor 2 (S1PR2) was identified as the GPCR receptor mediating the inhibitory effects of FAM19A5 on VSMC activation and migration.³⁸ It is worth noting that neither the predominant adipose expression nor binding of FAM19A5 to its receptor (S1PR2) has been independently confirmed.⁷⁹ These discrepancies warrant further clarification.

Fam19a5 is widely expressed in the mouse brain,²⁹ including hypothalamic proopiomelanocortin (POMC) and neuropeptide Y/agouti-related peptide neurons and microglia.³³ Inflammatory stimuli such as TNF-α (when delivered centrally) significantly upregulate Fam19a5 expression in the mouse hypothalamus.³³ Knocking down Fam19a5 expression partially reverses TNF-α-induced suppression of food intake, body weight loss, and inflammatory gene expression. Conversely, central delivery of recombinant FAM19A5 induces anorexia, body weight loss, and hyperthermia, along with elevations in inflammatory gene expression.³³ Reduced food intake is correlated with FAM19A5-induced activation of hypothalamic POMC neurons.³³ These results suggest a proinflammatory role of FAM19A5 in the brain. However, readers should interpret this study with caution since the source and purity of recombinant FAM19A5 were not indicated. If the protein was expressed and purified from bacteria, even a small residual level of LPS could potentially confound the results. Additionally, local concentration of FAM19A5 in the mouse cerebral spinal fluid is reported to be ~6 ng/mL;²⁴ thus the dose (1 μg/2 μL saline) injected into the right lateral ventricle and the effects observed may reflect a pharmacologic rather than a physiologic response. In this case, Fam19a5 KO mice could confirm a proinflammatory role and determine if central delivery of FAM19A5 reverses the KO phenotype.

In a study using LacZ reporter knock-in mice, Fam19a5 expression is upregulated by cryogenic-induced traumatic brain injury in a subset of neurons and oligodendrocyte precursor cells (OPCs).²⁶ Upregulated expression of Fam19a5 reporter is likely due to inflammation that accompanied brain injury, as inflammatory stimuli (e.g., TNF-α) increases Fam19a5 expression in the brain.³³ FAM19A5 thus may play a role in brain inflammation as previously suggested,³³ which is potentially relevant to certain human neuroinflammatory conditions.⁸⁰

Given the prominent expression of FAM19A5 in the CNS, Huang et al.²⁴ addressed the physiological consequences of disrupting Fam19a5 expression. Fam19a5 KO mice are hyperactive and exhibit greater anxiety/depressive-like behaviors, as they spend less time in the center of an open field compared to WT mice.²⁴ This increase in depressive-like behaviors is further revealed by sucrose preference and acute stress tests. In a sucrose preference test that measures degree of anhedonia (lack of interest in reward stimuli), Fam19a5 KO mice have a significantly lower preference for sucrose solution than WT controls.²⁴ In acute stress tests such as forced swim and tail suspension tests, Fam19a5 KO mice spend a greater percentage of time in the state of immobility compared to WT mice, a phenotype generally interpreted as an increased level of depression. Fam19a5 KO mice also have impaired hippocampus-dependent spatial learning and memory compared to WT mice, as determined by the Morris water maze test. The depression-like and memory deficit phenotypes seen in KO mice are attributed to reduced AMPA and NMDA receptor expression, lower glutamate release and neuronal activity in the hippocampus, and reduced dendritic spine density.²⁴ Interestingly, this study also showed that chronic stress (e.g., chronic restraint, forced swim, or social defeat) reduces FAM19A5 levels in the plasma and hippocampus of mice. AAV-mediated overexpression of human FAM19A5 in the mouse hippocampus ameliorates depressive-like behaviors caused by chronic stress.²⁴ Taken together, this study highlights FAM19A5 activity in the hippocampus modulating spatial learning and memory, as well as depressive-like behaviors. Notably, while all behavioral tests were conducted with genotypes blinded, the sex of mice was not indicated;²⁴ this biological variable is considered relevant and important in behavioral studies.^57–60 Intriguingly, Fam19a5 KO male and female mice in the study also have significantly reduced body weight, beginning at weaning (3 weeks of age), which persists until sexual maturity (8 weeks of age).²⁴ The body weight of mice older than 8 weeks is not reported. In addition, the authors also noted reduced absolute brain weight but increased brain/body weight ratio in Fam19a5 KO mice.²⁴ A change in both brain and body weight at a young age suggests possible developmental deficits in the KO animals, and these phenotypes could have contributed to the behavioral deficits seen in adult KO mice. Further, Fam19a5 is a reported target of Wnt9b/β-catenin in nephron progenitor cells.⁸¹ Thus, the possible developmental roles of FAM19A5 need to be further investigated and clarified. Developmental and behavioral alterations seen in mice lacking FAM19A5 may be relevant to humans, as serum levels of FAM19A5 are elevated in individuals with major depressive disorder⁸² and cognitive impairment.⁸³ However, we do not know whether increased circulating levels of FAM19A5 represent a potential compensatory response or merely a biomarker associated with disease. It is also unclear what peripheral tissue and cell type are responsible for the altered circulating levels of FAM19A5 in pathophysiological conditions.^80,82–84 In rare pediatric neurologic conditions (e.g., schizencephaly), micro-deletion of chromosome 22q13.32, which includes FAM19A5, has been reported;⁸⁵ the role of FAM19A5 in development of these conditions remains to be established.

FAM19A RECEPTORS

All chemokines bind to class A GPCRs of the rhodopsin-like family.⁸⁶ Thus, FAM19A proteins, distantly related to the chemokine MIP-1α, may act through GPCRs. Since many chemokine receptors induce calcium flux upon ligand binding,⁸⁷ a high-throughput calcium dynamic screening can be used to screen known chemokine receptors and structurally related orphan GPCRs for binding of FAM19A1. This assay reveals GPR1, a previously known receptor for chemerin,⁸⁸ as a candidate receptor for FAM19A1.²² Binding of FAM19A1 to GPR1 is further suggested by coimmunoprecipitation and flow cytometry analysis in cells expressing GPR1. In addition, binding with a synthetic peptide showed that the extracellular N-terminus of GPR1 interacts with FAM19A1 but not the related FAM19A5, further corroborating the specificity of the receptor–ligand interaction.²² These data suggest that GPR1 is a likely receptor for FAM19A1 (Figure 3A). Although knocking down the expression of GPR1 with siRNA significantly reduces binding of FAM19A1,²² CRISPR/Cas9-based gene deletion will provide a more definitive answer to whether GPR1 is required for FAM19A1 binding to target cells.

Figure 3.

FAM19A receptors. (A) FAM19A1 binds to G-protein-coupled receptor 1 (GPR1) with a dissociation constant (K_d) of 0.014 nM, and binding can be competed out by a synthetic peptide corresponding to the N-terminus portion of GPR1.²² (B) FAM19A4 binds to formyl peptide receptor 1 (FPR1) with a dissociation constant (K_d) of 0.447 nM, and binding can be competed out by either an unlabeled FAM19A4, an FPR1 antagonist (Boc-MLF), or the natural ligand for FPR1, N-formyl-met-leu-phe (fMLF).³⁷ (C) FAM19A5 binds to two putative receptors, formyl peptide receptor 2 (FPR2) and sphingosine-1-phosphate receptor 2 (S1PR2).^38,48 Binding of FAM19A5 to FPR2 can be competed out with an FPR2 antagonist; binding of FAM19A5 to S1PR2 can be competed out with either an unlabeled FAM19A5 or its natural ligand, sphingosine-1-phosphate (S1P).

Because FAM19A4 acts on macrophages to promote migration and phagocytosis, Wang et al.³⁷ searched the published literature for chemotactic receptors expressed in macrophages and found the GPCR FPR1, which is upregulated in macrophages polarized toward the M1 type and plays a role in macrophage chemotaxis.⁸⁹ An antagonist of FPR1 (Boc-MLF) can partially block the ability of FAM19A4 to promote macrophage phagocytosis and ROS release in vitro.³⁷ These data suggest that FPR1 is the putative receptor for FAM19A4 (Figure 3B). Future studies using cells from FPR1-deficient mice will help to further confirm whether FAM19A4 is a ligand for FPR1.

FAM19A5 can stimulate BMDM chemotaxis in a pertussistoxin-sensitive manner, suggesting it may act through a GPCR.⁴⁸ Because a previous study demonstrated that FPR1 is the receptor for FAM19A4,³⁷ Park et al. tested whether the related protein FPR2 could be a receptor for FAM19A5.⁴⁸ They found that expressing FPR2, but not FPR1, promotes FAM19A5-induced cell migration. Further, an antagonist of FPR2, but not FPR1, can block the ability of FAM19A5 to inhibit osteoclast formation in vitro. Importantly, FAM19A5 fails to elicit chemotaxis and osteoclastogenesis in BMDMs obtained from FPR2 knockout mice.⁴⁸ These data suggest that FPR2 is the putative receptor for FAM19A5 (Figure 3C). However, the binding affinity of FAM19A5 to FPR2 has not been established.

FAM19A5 also acts on VSMCs.³⁸ To identify the receptor on VSMCs that mediates the biological activity of FAM19A5, public databases were searched for all class A GPCRs expressed in VSMCs.³⁸ Of the 12 candidate GPCRs, 6 were implicated in VSMC proliferation and neointima formation following vascular injury. Among these six GPCRs, four (S1PR1, S1PR3, GRP180, F2R/PAR1) promote VSMC proliferation, migration, and neointima formation^90–93 and two (S1PR2 and GPR176) negatively regulate neointima formation.^94,95 siRNA-mediated silencing of F2R, GPR180, and S1PR1 has no effect on FAM19A5-induced suppression of VSMC proliferation; in contrast, silencing of S1PR2 partially blocks the effect of FAM19A5 on VSMC proliferation.³⁸ S1PR2-specific antagonist similarly blocks effects of FAM19A5 on VSMC proliferation and migration.³⁸ Coimmunoprecipitation and flow cytometry analysis further suggest binding of FAM19A5 to S1PR2. While these data support the notion that S1PR2 is a putative receptor for FAM19A5 (Figure 3C), the findings have recently been challenged and disputed.⁷⁹ Future studies using VSMC derived from S1PR2 knockout mice are needed to help resolve the issue and clarify whether FAM19A5 can bind two different GPCRs (FPR2 and S1PR2).

Neurexin receptors bind diverse ligands.⁴⁰ Immunopurification of endogenous neurexin-1 (NRXN1) from mouse brain followed by mass spectrometry analysis identified FAM19A1 and FAM19A2 as two novel proteins coimmunoprecipitated with NRXN1.³⁶ In coexpression studies, FAM19A1–4 form disulfide-bonded complexes with all neurexin splice variants. FAM19A/NRXN1 complex formation occurs only in the ER/Golgi secretory pathway; mixing separately purified proteins fails to recapitulate complex formation. FAM19A1 forms covalent bonds with two cysteine residues in the cysteine loop domain located at the juxta-membrane stalk region of NRXN1. Overexpressing FAM19A1 in hippocampal cultures reduces O-glycosylation of NRXN1 and suppresses its heparan sulfate modification.³⁶ FAM19As may therefore be cell-type-specific regulators of neurexin modifications instead of traditional secreted ligands that bind neurexins.³⁶ Considering these findings, it would be informative to determine whether posttranslational modifications of neurexins are altered in each of the FAM19A single-gene knockout mice generated to date. Since FAM19A1–4 could form complexes with neurexins when coexpressed in cells, this redundancy may require deletion of two or more of the FAM19A family members to observe potential changes in the posttranslational modifications of neurexins.

PERSPECTIVE AND FUTURE DIRECTION

Great strides have been made in unraveling the biological functions of the novel FAM19A family of neurokines. Analyses of single-gene KO mouse or zebrafish models reveal nonredundant roles for each FAM19A family member in modulating locomotor activity, anxiety- and depressive-like behaviors, social communication, repetitive behaviors, spatial learning and memory, and somatosensory functions. Given the region-specific expression of each FAM19A in the central and peripheral nervous system, we anticipate that additional behavioral and functional deficits will be uncovered in future studies using loss-of-function animal models. Except for FAM19A4,^51,63 the mechanism of action and cell targets (neurons, astrocytes, microglia, oligodendrocytes) for FAM19As in the central and peripheral nervous system remain largely undefined. Pinpointing neurocircuits and signaling pathways controlled by FAM19A in different parts of the nervous system, as well as different neuronal cell targets, will be the next challenge. For example, whole-cell electrophysiological recordings could be used to establish whether different FAM19A family members act on excitatory or inhibitory neurons. Genetically encoded calcium indicator,⁹⁶ along with cell-type specific markers, can be used in neuronal slice cultures to interrogate which cells are activated by which FAM19A family members. Cell-type-selective inactivation of FAM19A and conditional KO models will need to be generated to uncover central vs peripheral roles of FAM19A and the possible source of altered circulating levels in disease contexts. Since there are functional clusters of cells that express a combination of two or more Fam19a genes,²⁹ double KO mice would help resolve potential redundancy as well as additive or synergistic effects on behaviors due to multiple gene deletions. Although much remains to be discovered, studies of gain- and loss-of-function animal models have provided a critical foundation and physiological context for future detailed mechanistic studies.

ABBREVIATIONS

AMPA

α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid

BMDM

bone marrow-derived macrophage

CCL3

C–C motif chemokine ligand 3

CNS

central nervous system

CBP

CREB binding protein

CREB

cAMP-response element binding protein

CRH

corticotropin releasing hormone

DRG

dorsal root ganglion

Erk1

extracellular signal-regulated kinase 1

FABP4

fatty acid binding protein 4

FAM19A

family with sequence similarity 19, member A

FRP1

formyl peptide receptor 1

GPR1

G-protein-coupled receptor 1

IL-6

interleukin-6

KO

knockout

LPS

lipopolysaccharide

MIP-1α

macrophage inflammatory protein 1α

MT1

melatonin receptor subtype 1

NE

norepinephrine

NF1

neurofibromin

NMDA

N-methyl-d-aspartic acid

OPC

oligodendrocyte precursor cell

PVN

paraventricular nucleus

Rac1

Ras-related C3 botulinum toxin substrate 1

ROCK

Rho-associated protein kinase

ROS

reactive oxygen species

SAM2

samdori 2

STAT3

signal transducer and activator of transcription 3

TNF-α

tumor necrosis factor α

VSMC

vascular smooth muscle cell

WT

wild type