Human GLS2 gene is located on chromosome 12 (Aledo et al., 2000). It is composed of 18 exons and spans approximately 18 kb (Pérez-Gómez et al., 2003). The gene resides on the minus strand. It starts at 56470944 and ends at 56488414 from pter (NCBI, Gene ID 27165).

Two GLS2 transcripts coding for functional proteins have been described: a long canonical one containing all 18 exons (GAB), cloned from a human breast cancer cell line (Gómez-Fabre et al., 2000) and a short variant (LGA), first characterized in rat liver (Smith and Watford, 1990; Chung-Bok et al., 1997) and lacking the first exon (Martèn-Rufián et al., 2012). LGA transcript arises by a combination of two mechanisms of transcriptional regulation: alternative transcription initiation and alternative promoter. Its transcription start site is located at 3-end of the first intron of GLS2 gene. Other non-coding transcripts, containing premature stop codons, have been isolated (Martèn-Rufián et al., 2012).
The molecular basis for GLS2 regulation is now starting to be uncovered. GLS2 has been identified as a transcriptional target of tumor suppressor TP53 that mediates its new revealed functions in tumor metabolism and antioxidant defense, under both non-stressed and stressed conditions. This tumor suppressor directly associates with response elements in the GLS2 promoter (Hu et al., 2010; Suzuki et al., 2010). Cells with heightened GLS2 levels showed increased production of glutamate and alpha-ketoglutarate, which resulted in enhanced oxydative phosphorylation, higher GSH/GSSG ratios and decreased reactive oxygen species (ROS) levels, which provided protection against ROS induced apoptosis (Hu et al., 2010; Suzuki et al., 2010). Two other transcription factors belonging to the TP53 family, TP73 and TP63, also drive the expression of GLS2 during neuronal differentiation of neuroblastoma cells after induction with retinoic acid (Velletri et al., 2013) and during epidermal differentiation and in cancer cells exposed to oxidative stress (Giacobbe et al., 2013), respectively. The GLS2 downregulation observed in liver and colon cancer cell lines and in hepatocellular carcinoma has been attributed to its promoter hypermethylation, a mechanism that has been proposed as a marker to identify novel tumor suppressor genes. Chemical demethylation treatment increased the GLS2 mRNA levels in these cells (Zhang et al., 2013; Liu et al., 2014). This mechanism is also responsible for the GLS2 silencing detected in highly malignant glioblastomas, and occurs regardless of their TP53 status (Szeliga et al., 2016). Other factors that regulate GLS2 expression are the MYCN oncoprotein (v-myc avian myelocytomatosis viral oncogene neuroblastoma derived homolog, encoded by MYCN gene) and the NR5A2 (nuclear receptor subfamily 5 group A member 2), also known as LRH-1 (liver receptor homolog 1). MYCN, an essential MYC family member, directly activates GLS2 transcription in MYCN-amplified neuroblastoma cells. Depletion of MYCN expression by short hairpin RNA caused a pronounced decrease in GLS2 (but not GLS) levels in these cells. Conversely, inhibition of TP53 had no effect on GLS2 induction (Xiao et al., 2015). In hepatoma cells, GLS2 is subjected to direct transcriptional regulation by NR5A2. Silencing of NR5A2 mediated by small interfering RNA reduced GLS2 transcript and protein levels (Xu et al., 2016). Long non-coding RNA (lncRNA) and microRNAs (miRNAs) are also implicated in GLS2 regulation. The lncRNA urothelial carcinoma-associated 1 ( UCA1) regulates GLS2 expression levels through interfering with MIR16-1, which binds to the 3-UTR (untranslated region) of GLS2 mRNA (Li et al., 2016).

At least one pseudogene has been reported for GLS2 (GenBank: AF110329.1).

Purification: the first attempts to purify liver-type glutaminase from rat liver only yielded partial purifications: 15-fold (Huang and Knox, 1976) and 60-fold (Patel and McGivan, 1984). Unlike kidney-type glutaminase (KGA), the liver isoenzyme does not polymerize in the presence of phosphate-borate buffer (Huang and Knox, 1976). This different behaviour, together with its greater instability in tissue homogenates and diluted preparations, made this isoenzyme more difficult to purify than KGA. In fact, its purification to near homogeneity was not achieved until almost two decades later than KGA purification. Heini et al. (1987) purified rat LGA 400-fold, while Smith and Watford (1988) reported a 600-fold purification, with specific activities of 30-50 μmol/min per mg protein. Human GAB isoenzyme has been expressed as a recombinant protein in baculovirus system and purified by affinity chromatography with a specific activity of 18 μmol/min per mg protein (Campos-Sandoval et al., 2007).
Structure: human GAB transcript (ORF: 1809 nt) codes a 602-residues protein, with a predicted molecular mass of 66.3 kDa. Human LGA transcript (ORF: 1698 nt) codes a 565-residues protein, with a predicted molecular mass of 62.5 kDa. The precursor of human LGA isoform lacks the first 61 residues of GAB precursor (coded by exon 1), but it displays an additional extension of 24 novel residues at the N-terminus coded by an alternative first exon. This extension and the first six amino acids coded by exon 2 are not present in its rat and mouse counterparts (Martèn-Rufián et al., 2012). GLS2 contains a central glutaminase domain of approximately 300 residues, which belongs to the beta lactamase/transpeptidase-like superfamily, and three ankyrin repeats at the C-terminal region (Pasquali et al., 2017). The ankyrin repeats are protein-protein interaction modules of 33 residues that have been found in many important proteins such as transcriptional factors, cell cycle regulators, cytoskeletal organizers, etc. (Sedgwick and Smerdon, 1999; Mosavi et al., 2004). At its C-terminal end there is a consensus sequence of four residues required for specific interaction with PDZ (postsynaptic density protein, disc large, zona occludens) proteins (Olalla et al., 2001). The N-terminal end (first 14 residues) of GAB precursor contains a putative mitochondrial import presequence (Gómez-Fabre et al., 2000). It is worth mentioning the presence of a consensus LXXLL motif of interaction with nuclear receptors at N-terminal region of GLS2 (Olalla et al., 2002). It is in the regions involved with organelle targeting (exon 1) and protein-protein interactions (exon 18) that the main differences between GAB and kydney-type glutaminase reside (Pérez-Gómez et al., 2003). Studies on processing and molecular structure of native GLS2 protein are lacking. An apparent subunit molecular mass of 57-58 kDa for the rat liver isoform was determined by denaturing gel electrophoresis of purified protein (Heini et al., 1987; Smith and Watford, 1988). Sequencing by Edman degradation of the mature form of human GAB expressed in baculovirus system showed cleavage between amino acids 38-39 and 39-40 of the deduced protein sequence. These cleavages are consistent with known substrate sites for the mitochondrial processing peptidase (MPP), having an Arg residue at position -2 or -3. Nevertheless, processing in baculovirus system may somehow differ from the native processing in mammals (Campos-Sandoval et al., 2007). The molecular mass (Mr) of native GLS2 has not been determined accurately. Smith and Watford (1988) reported an Mr of ≥ 300000 from HPLC gel filtration but obtained a value of 162000 by sucrose gradient centrifugation, regardless of phosphate concentration (5 or 100 mM).
Kinetic properties: the distinct kinetic behavior of mammalian kydney-type and liver-type isoenzymes was first noted by Krebs (1935). In contrast with kydney-type isoenzyme, the rat liver-type glutaminase showed a lower dependence on the activator inorganic phosphate (Pi), lower affinity for the substrate glutamine, lack of inhibition by glutamate (up to 50 mM) and a requirement for ammonia as an obligatory activator (Verhoeven et al., 1983; Patel and McGivan, 1984; Smith and Watford, 1988). The only human GLS2 protein characterized in a purified form is the recombinant GAB expressed in insect cells. It showed an allosteric behavior (Hill index of 2.7) with low affinity for glutamine (S0.5 values of 32 and 64 mM for high (150 mM) and low (5 mM) Pi, respectively), and low dependence for Pi as expected for a GLS2 isoenzyme. Surprisingly, GAB was inhibited by glutamate, a characteristic only shown by GLS isoforms, with an IC50 value of 50 mM at low Pi concentrations (5 mM) and suboptimal glutamine concentration (20 mM), and scarcely activated by ammonia (Campos-Sandoval et al., 2007).
Post-translational modifications: several acetylated and succinylated lysine residues have been identified in GLS2 by large-scale proteomic approaches (Rardin et al., 2013; Park et al., 2013).

The classical pattern of glutaminase expression first established that GLS2 isoforms were restricted to postnatal liver, while GLS isoforms were widely distributed in most nonhepatic tissues (Curthoys and Watford, 1995). Recent findings have extended the range of GLS2 expression to other extrahepatic tissues such as brain, pancreas, cells of immune system and tumor cells (Aledo et al., 2000; Castell et al., 2004; Gómez-Fabre et al., 2000; Pérez-Gómez et al., 2005). Experimental evidences now support that both glutaminase genes are coexpressed in some tissues as well as in many cell types. Thus, GLS2 co-localized with KGA in numerous cells throughout the brain (Olalla et al., 2002; Cardona et al., 2015). Both GLS2 and GLS transcripts were also found to be co-expressed in liver and brain of rat, mouse and human (Martèn-Rufián et al., 2012). Simultaneous expression of GLS and GLS2 isoforms were also confirmed for several human cancer cells at the transcriptome and proteome levels (Turner and McGivan, 2003; Pérez-Gómez et al., 2005). In different types of cancer, such as bladder, colon and lung cancer, GLS2 is considerably overexpressed compared with normal tissues (Saha et al., 2019).

Although glutaminase has long been considered predominantly a mitochondrial enzyme (Curthoys and Watford, 1995), a differential localization for glutaminase was found in neurons using isoform-specific antibodies (Campos et al., 2003): while KGA was present in mitochondria, a Gls2-encoded isoform - most probably GAB - was found in nuclei where it was catalytically active (Olalla et al., 2002).

GLS2 (E.C. 3.5.1.2.) catalyzes the hydrolytic deamidation of L-glutamine to form L-glutamate and ammonium, the first step of glutaminolysis. In liver, GLS2 reaction provides substrates for gluconeogenesis and urea synthesis (Watford, 1993). A role of GLS2 in neuronal differentiation has been recently reported (Velletri et al., 2013). In some cancer cells, GLS2 may play a role as tumor suppressor, in opposition to GLS, regulated by oncogenes and associated to tumorigenesis. GLS2 overexpression in glioblastoma cell lines caused a reversion of their transformed phenotype (Szeliga et al., 2009). This is in agreement with the loss of GLS2 expression in hepatocellular carcinomas (Suzuki et al., 2010) and brain tumors (Szeliga et al., 2005). However, this behavior of GLS2 is not universal, as there are some types of cancer - cervical and lung cancers, NMYC-amplified neuroblastoma - where GLS2 expression is upregulated and this upregulation is associated with therapeutic resistance and poor clinical outcomes (Xiang et al., 2013; Xiang et al., 2015; Saha et al., 2019).
Interacting partners: the first protein-interacting partners of GLS2 were discovered by two-hybrid genetic screening of a human brain cDNA library, using the C-terminal region of GLS2 as bait. Two PDZ domain-containing proteins were isolated: SNTA1 (alpha 1-syntrophin) and TAX1BP3 (Tax1-binding protein 3, also known as Tax-interacting protein 1 (TIP-1) or glutaminase-interacting protein (GIP)). The C-terminal end of human GLS2, -ESMV, matches the consensus sequence X-Ser/Thr-X-Val required for interaction with PDZ proteins (Olalla et al., 2001). A dissociation constant of 1.66 μM was determined for the GLS2-GIP binding, which indicates a moderate affinity suitable for regulatory functions (Banerjee et al., 2008). Two recent findings have revealed the important role that GLS2 plays in tumor suppression through its interactions with other proteins in a glutaminase activity independent manner. GLS2 binds through its C-terminal region to RAC1 (Rac family small GTPase 1), a critical promoter of metastasis frequently activated in several types of cancer, and inhibits its activation by guanine-nucleotide exchange factors (GEFs). Thus, as a direct target of TP53, GLS2 mediates TP 53s function in metastasis suppression (Zhang et al., 2016). GLS2 also binds to and stabilizes DICER1, a key component of the microRNAs processing machinery, promoting MIR34A maturation. Upregulation of this miRNA represses metastasis in hepatoma cells through expression inhibition of SNAI1 (snail family transcriptional repressor 1), a transcriptional repressor of cadherin 1 ( CDH1, also known as E-cadherin) (Kuo et al., 2016).