Lactate as a determinant of neuronal excitability, neuroenergetics and beyond

Over the last decades, lactate has emerged as important energy substrate for the brain fueling of neurons. A growing body of evidence now indicates that it is also a signaling molecule modulating neuronal excitability and activity as well as brain functions. In this review, we will briefly summarize how different cell types produce and release lactate. We will further describe different signaling mechanisms allowing lactate to fine-tune neuronal excitability and activity, and will finally discuss how these mechanisms could cooperate to modulate neuro-energetics and higher order brain functions both in physiological and pathological conditions.


Introduction
Lactate, first isolated in 1780 from sour milk by the Swedish chemist Carl Wilhelm Scheele (Oesper, 1931), is an end-product of carbohydrate fermentation occurring in some bacteria and yeasts under anaerobic conditions. It is also produced by animal cells when oxygen availability becomes limited, such as during intense muscle exercise (Li et al., 2022) or in pathological conditions like stroke (Henriksen et al., 1992), cancer (Hayes et al., 2021) and epilepsy (Slais et al., 2008). During anaerobic glycolysis, pyruvate is indeed converted into lactate instead of entering the Krebs cycle and producing ATP, the energy fuel of the cell. In the 1920s, Otto Warburg discovered that the "fermentation" of glucose into lactate occurred in cancer cells despite the presence of oxygen (Warburg et al., 1927). Since then, this metabolic pathway, referred to as aerobic glycolysis or "Warburg effect", has also been described under physiological conditions, notably in proliferating cells (Vander Heiden et al., 2009). More than a century ago, lactate was considered as a useless waste product, potentially toxic and responsible for muscle fatigue, soreness and rigor mortis (Schurr, 2014;Hall et al., 2016). Despite evidence that some organs, including the heart and the brain, are able to consume an excess of lactate, this process was only considered as a means to remove lactate from the blood (Schurr, 2014). This "bad reputation" slowly began to change in the late 1920s when the Coris showed that d-lactate (former name of L-lactate), "when fed by the mouth or injected subcutaneously, leads to glycogen deposition in the liver" (Cori and Cori, 1929). Indeed, blood lactate recycled through gluconeogenesis and glycogenesis into liver glycogen can in turn lead to blood glucose, muscle glycogen and blood lactate again through glycolysis (see the Cori cycle).
This review therefore aims at providing an overview of the cellular and molecular mechanisms underlying the modulating impact of lactate on neuronal excitability and activity in cortical areas and integrated brain functions. Before getting to the heart of integrated brain functions, we will describe the sources of lactate as well as the mechanisms by which lactate modulates neuronal excitability and activity. Finally, we will discuss how these mechanisms contribute to integrated functions by presenting some examples: both in physiological conditions, including neurovascular and neurometabolic couplings (NVC and NMC) and learning and memory, and in pathological conditions.
Lactate is also produced by skeletal muscles during physical exercise (Dalsgaard et al., 2004;Morland et al., 2017;Li et al., 2022), and in particular by the fast-twitch glycolytic fibers, which shuttle lactate to the slow-twitch oxidative fibers (Brooks, 2009). Skeletal muscle lactate is also released in the blood via MCT1 and MCT4 (Manning Fox et al., 2000;Bonen et al., 2000). Blood-borne lactate can cross the blood-brain barrier via MCT1 (Bergersen et al., 2002;Pierre and Pellerin, 2005;Chiry et al., 2006), which is present at both the luminal and the abluminal membranes of endothelial cells (Zlokovic, 2008). When its plasma concentration exceeds that of the brain extracellular space (Machler et al., 2016;Carrard et al., 2018;Roumes et al., 2021b), following an intense physical exercise for example (Dalsgaard et al., 2004;Morland et al., 2017;Li et al., 2022), it can be transported along its gradient and reach brain parenchyma.
In summary, the main sources of lactate in the brain parenchyma are astrocytes and blood.

How lactate modulates neuronal excitability
Once extracellular lactate is increased in the cortical parenchyma, it may affect neuronal excitability and activity mainly through three nonexclusive mechanisms that are summarized below: i) lactate receptor, ii) N-methyl-D-aspartate (NMDA) receptors and iii) ATP-sensitive K + (K ATP ) channels ( Fig. 1, points 5, 8, 10).

Lactate receptor
Hydroxycarboxylic acid receptor 1 (HCAR1, previously termed G protein receptor 81, GPR81) is a G i protein-coupled receptor that is activated by the metabolically relevant enantiomer L-lactate with an EC 50 of 1-7 mM, and less efficiently by its partial agonist D-lactate, but not by pyruvate (Cai et al., 2008;Liu et al., 2009;Ahmed et al., 2010) (Table 1). HCAR1 transcripts are mostly expressed in white adipose tissue (Regard et al., 2008) but are also observed in hippocampus and neocortex, although to a much lesser extent (Lauritzen et al., 2014;Briquet et al., 2022). HCAR1 mRNAs are barely detectable in mouse cortical neurons by either single-cell RNAseq (Zeisel et al., 2015;Tasic et al., 2016) or RNAscope TM (Briquet et al., 2022), indicating low expression at the cellular level. Consistently, because of the lack of specific antibodies (de Castro Abrantes et al., 2019), its cellular expression at the protein level has been indirectly reported in pial fibroblast-like and pericyte-like cells as well as in neurons using mice expressing the monomeric red fluorescent protein under the Hcar1 promoter (Morland et al., 2017;de Castro Abrantes et al., 2019;Briquet et al., 2022). When applied on primary cultures of embryonic cortical neurons, Llactate decreased calcium transients, assessed as a surrogate of spiking activity, with an IC 50 of 4.2 mM (Bozzo et al., 2013). D-lactate, which is much less efficiently transported than L-lactate (Nedergaard and Goldman, 1993) and poorly metabolized by mammalian cells (Cori and Cori, 1929;Ewaschuk et al., 2005), decreased cortical neuron spiking activity with an efficacy (IC 50 = 4.6 mM) similar to that of L-lactate ( Fig. 1, point 5 and Table 1). The HCAR1 specific full agonist 3,5-dyhydroxybenzoic acid , but not pyruvate (Table 1) or glucose, mimicked this effect suggesting that HCAR1 was indeed responsible for this modulation (Bozzo et al., 2013). The involvement of this receptor in decreased excitability was confirmed by showing that L-lactate and HCAR1 agonists lowered the spontaneous calcium activity of embryonic cortical neurons from wild-type but not from HCAR1 knockout mice (de Castro Abrantes et al., 2019). Using whole-cell patch-clamp recording in hippocampal slices from young male rats, HCAR1 agonists were also found to decrease the excitability of CA1 neurons through a G αi subunit independently of MCTs blockade (Herrera-Lopez and Galvan, 2018). Similar inhibitory effects of HCAR1 activation on neuronal activity were confirmed in embryonic cortical neurons and extended to the presynaptic regulation of glutamate release (de Castro Abrantes et al., 2019). Whole-cell patch-clamp recording of dentate gyrus slices from young rodents revealed that HCAR1 activation decreased the excitability of granule neurons with similar pre-and post-synaptic mechanisms, while this modulation was not detected in mice lacking HCAR1 (Briquet et al., 2022). Likewise, the activity of cortical neurons recorded from epileptic patient brain tissues was decreased by an HCAR1 agonist as revealed by patch-clamp recording and Ca 2+ imaging (Briquet et al., 2022).
In summary, the activation of the lactate receptor HCAR1 decreases the activity and excitability of cortical neurons via pre-and postsynaptic mechanisms (Table 1). Further studies would be necessary to generalize this effect of lactate. Due to the lack of specific antibodies, the precise expression and localization of HCAR1 remains to be confirmed in neurons (de Castro Abrantes et al., 2019;Briquet et al., 2022). However, the use of HCAR1 knockout mice attests for the functional expression of HCAR1 in cortical neurons. Besides, the use of whole-cell patch-clamp recording (Herrera-Lopez and Galvan, 2018;de Castro Abrantes et al., 2019;Briquet et al., 2022), may have altered intracellular metabolism (Horn and Marty, 1988) and thereby the relative contribution of lactate receptors and metabolism on neuronal excitability. Investigating neuronal activity with Ca 2+ imaging is an alternative approach to preserve intracellular metabolism (Bozzo et al., 2013). The role of HCAR1 should be considered in a condition-dependent manner such as in embryonic neurons (Bozzo et al., 2013;de Castro Abrantes et al., 2019) or in epileptic tissues (Briquet et al., 2022).

Modulation of NMDA receptors by lactate signaling
NMDA receptors represent a subtype of glutamate-gated ion channels with unique properties distinguishing them from other ionotropic glutamate receptors. In particular they display a high Ca 2+ permeability (MacDermott et al., 1986), a voltage-dependent Mg 2+ block (Nowak et al., 1984), a slow gating kinetics (Lester et al., 1990) and are specifically activated by both glutamate and glycine (Johnson and Ascher, 1987) or D-serine (Mothet et al., 2000) as co-agonists.
In primary cultures of mouse neocortical neurons, co-application of glutamate and a saturating concentration of glycine (Priestley et al., 1995) induced NMDA currents and the associated intracellular Ca 2+ increase (MacDermott et al., 1986). Both responses were potentiated in the presence of 10 mM L-Lactate (Yang et al., 2014;Jourdain et al., 2018). Similar concentrations of D-lactate or pyruvate, or even an isoenergetic glucose concentration did not increase (Table 1), but instead decreased this calcium response (Yang et al., 2014;Jourdain et al., 2018). This lack of enhancement by pyruvate and isoenergetic glucose ruled out the possibility that the effect of L-lactate on NMDA-induced Ca 2+ response was mediated by energy metabolism. The decreased Ca 2+ response by D-lactate could involve HCAR1 receptor, as shown in cultured cortical neurons (Bozzo et al., 2013;de Castro Abrantes et al., 2019). However, the opposite effects of lactate and pyruvate suggest the contribution of redox state modulation (Fig. 1, point 8). Indeed, exogenous application of 4 mM NADH, which is generated by LDH during Llactate oxidation (Williamson et al., 1967), also enhanced NMDAdependent intracellular Ca 2+ increase (Yang et al., 2014). Consistently, LDH inhibition by stiripentol (Sada et al., 2015) prevented the enhancing effect of L-lactate (Jourdain et al., 2018).
The potentiating effect of L-lactate, but not the basal Ca 2+ response induced by glutamate/glycine co-application, was found to recruit NR2B-containing NMDA receptors (Jourdain et al., 2018). Supporting a redox-state effect of lactate on NMDA receptors, dithiothreitol, a reducing agent which potentiates NMDA currents (Sullivan et al., 1994;Kohr et al., 1994), also enhanced the intracellular Ca 2+ increase induced by glutamate and glycine in a NR2B-dependent manner (Jourdain et al., 2018). Conversely, the oxidizing agent 5,5′-dithiobis(2-nitrobenzoic acid) prevented the effect of L-lactate (Jourdain et al., 2018). In addition to this mechanism, L-Lactate may also indirectly contribute to glutamatergic transmission. Indeed, L-lactate has been shown to promote extracellular prostaglandin E2 (PGE2) accumulation through prostaglandin transporter (PGT) modulation (Gordon et al., 2008). Given that PGE2 is a retrograde lipid messenger facilitating glutamatergic transmission (Sang et al., 2005), lactate may also indirectly potentiate glutamatergic transmission (Fig. 1, point 9), and thus NMDA receptor activity.
In summary, the enhancing effect of L-lactate on neuronal activity via NMDA receptors appears to mostly involve the redox regulation of NR2B-containing receptors by NADH produced intracellularly by LDH. This process thus requires L-lactate uptake (Fig. 1, point 8 and Table 1) and potentially the glutamatergic transmission via PGE2 at the presynaptic terminals (Fig. 1, point 9).

The Janus regulation of K ATP channels by lactate
Before addressing whether and how L-lactate influences energy metabolism-dependent homeostatic regulation, we will introduce the key players of this regulation, the K ATP channels, and will discuss how this homeostatic regulation modulates neuronal excitability and activity. Finally we will report on how L-lactate may either close or open K ATP channels.
Indeed, in adult rat entorhinal cortical slices, metabolically sensitive slow wave oscillations have been recorded with sharp electrodes in both excitatory neurons and inhibitory interneurons (Cunningham et al., 2006). The frequency of these oscillations, which consisted in up (depolarization) and down (hyperpolarization) phases, was decreased by lowering the extracellular glucose from 10 mM, a classically used ex vivo concentration, to a physiologically relevant concentration of 2.5 mM (Silver and Erecinska, 1994;Hu and Wilson, 1997b). Down states were prolonged in 2.5 mM glucose, whereas the up states duration, relying on excitatory recurrent activity, remained unaffected. This observation indicated that only the down phase was metabolically sensitive (Cunningham et al., 2006). It also suggested that the decreased energy charge (i.e., ATP/ADP ratio) of neurons during up states, which involved highly energy-demanding excitatory transmission (Attwell and Laughlin, 2001;Tantama et al., 2013;Baeza-Lehnert et al., 2019), drove the transition to the down phase by activation of K ATP channels. Consistently, up states and down states were mimicked by K ATP blocker and opener, respectively. Moreover, intracellular recording of single neurons with a high ATP concentration in the recording pipette, to block K ATP channels, prolonged the up phase and suppressed the hyperpolarization of the down phase in the recorded neurons, without affecting population activity. Hence, this pioneering work disclosed an energy metabolismdependent homeostatic regulation of cortical activity (Cunningham et al., 2006).
Lactate-induced neuronal depolarisation via K ATP channels closure: The importance of L-lactate, LDH, the downstream pyruvate and K ATP channels has been shown in the control of neuronal excitability in different brain areas (Sada et al., 2015). Using whole-cell patch-clamp recording in hippocampal slices, CA1 pyramidal cells were hyperpolarized by glucose depletion or showed impaired L-lactate metabolism by LDH inhibition (Sada et al., 2015). As discussed below, it was not the case of fast spiking interneurons despite a functional expression of K ATP channels. This hyperpolarization induced by LDH inhibition was reversed by extracellular pyruvate, and was accompanied by a decrease in membrane resistance, indicative of hyperpolarizing conductance activation and decreased excitability. Consistent with a neuron-type specific effect of LDH inhibition, excitatory postsynaptic currents were more strongly reduced than inhibitory postsynaptic currents. Furthermore, paired whole-cell and cell-attached recording between CA1 pyramidal cells and nearby astrocytes were used to selectively inhibit LDH in astrocytes upon rupture of the cell-attached membrane patch. This treatment induced a progressive hyperpolarisation in pyramidal cells and was prevented by extracellular L-lactate, thereby supporting the ANLS (Sada et al., 2015). The lack of effect in fast spiking interneurons was attributed to a missing link between downstream LDH metabolite and K ATP channels. Of note, intracellular ATP delivered through the patch pipette did not revert the hyperpolarization induced by LDH inhibition in neurons of the basal ganglia subthalamic nucleus.
In contrast with neurons of the entorhinal cortex (Cunningham et al., 2006), the evoked firing of single neurons from the barrel cortex, which were recorded in slices of juvenile rodents using the perforated patch configuration that preserves intracellular metabolism (Horn and Marty, 1988), was found to be insensitive to a decrease in glucose concentration from 10 mM to 2.5 mM. Under 2.5 mM extracellular glucose, L-lactate dose-dependently enhanced the spiking activity of both pyramidal cells and cortical GABAergic interneurons (Karagiannis et al., 2021). This enhancing effect was observed at a concentration of L-lactate up to 15 mM, an isoenergetic condition with 10 mM glucose, which doubled the firing rate. 15 mM pyruvate induced a similar enhancing effect, ruling out the involvement of HCAR1 (Ahmed et al., 2010) and a redox-state dependence (Table 1). Lactate-induced increase in firing rate was abolished by MCT blockade implying the facilitated transport of Llactate for its enhancing effect on spiking activity. As evidenced by NADH intrinsic fluorescence imaging (Chance et al., 1962), cortical neurons were found to uptake and oxidize L-lactate by LDH (Williamson et al., 1967;Karagiannis et al., 2021) (Table 1). Consistent with a high oxidative metabolism of cortical neurons (Itoh et al., 2003;Gulyas et al., 2006), the majority of the produced ATP was found to massively derive from oxidative metabolism and poorly from glycolysis (Karagiannis et al., 2021) (Fig. 1, point 7), as shown by the use of an intracellular fluorescent ATP biosensor (Imamura et al., 2009) and metabolic inhibitors. These observations suggested that the L-lactate enhancing effect was mediated by the downstream oxidative metabolism of pyruvate in mitochondria to produce ATP and blockade of K ATP channels (Fig. 1,  point 10). Indeed, the firing-rate enhancement by L-lactate was suppressed by a K ATP channel opener, while this effect was reversed by a K ATP channel blocker. L-lactate occluded the effect of the blocker, indicating the closure of all K ATP channels. Furthermore, cortical neurons of Kir6.2 knockout mice were insensitive to L-lactate, thereby implying closure of K ATP channels in lactate sensing.
Lactate-induced neuroprotection via K ATP channels opening. In primary cultures of cortical neurons, application of glutamate in a Mg 2+ -free medium (Jourdain et al., 2016), or co-application with glycine at saturating concentrations, induced a massive NMDA-dependent Ca 2+ entry followed by an irreversible cell swelling (Jourdain et al., 2018), as imaged by digital holography microscopy (Jourdain et al., 2011), which preceded cell death. L-lactate dose-dependently protected against this NMDA-receptor-dependent excitotoxicity (Jourdain et al., 2016;Jourdain et al., 2018). This neuroprotection was mimicked with a similar concentration of pyruvate, but not by 10 mM D-lactate (Jourdain et al., 2016;Jourdain et al., 2018), suggesting that the effect of L-lactate was independent of redox state and HCAR1 receptor (Ahmed et al., 2010) ( Table 1). The absence of effect using an isoenergetic condition with 5 mM glucose indicated that neuroprotection did not rely on glucose energy metabolism (Jourdain et al., 2016). The observations that i) Llactate is preferred over glucose as an energy substrate in cultured neurons (Bouzier-Sore et al., 2003;Itoh et al., 2003), ii) L-lactate is converted into pyruvate by LDH, and iii) L-lactate and pyruvate have similar neuroprotective effects, suggested a neuroprotection mediated by the oxidative metabolism of pyruvate in mitochondria to produce ATP. Indeed, L-lactate-induced neuroprotection was abolished by the blockade of the mitochondrial pyruvate carrier (Jourdain et al., 2016). Furthermore, extracellular application of ATPγS, a non metabolisable ATP analogue and a stable purinergic receptor, but not of an adenosine receptor agonist, mimicked the neuroprotective effect of L-lactate, suggesting that it was not mediated by ATP consumption but rather by the release of ATP and the activation of extracellular purinergic receptors (Jourdain et al., 2016). The neuroprotection conferred by Llactate was abolished by the blockade of pannexins, which act as ATPreleasing channels (Bao et al., 2004;Whyte-Fagundes and Zoidl, 2018), as well as by the presence of an extracellular ATP-degrading enzyme, thereby ruling out the involvement of adenosine receptors (Jourdain et al., 2016;Jourdain et al., 2018). Molecular observations and pharmacological manipulations respectively identified P2Y2 receptor and its downstream phosphatidylinositol 3'-kinase (PI3K) pathway, as the purinergic receptor and its signalling cascade responsible for L-lactate-induced neuroprotection (Jourdain et al., 2016). Since the activation of K ATP channels is neuroprotective by reducing neuronal activity and energy demand (Soundarapandian et al., 2007) and the PI3K pathway has been shown to open these channels (Mirshamsi et al., 2004;Plum et al., 2006), their contribution to L-lactate-induced neuroprotection was evaluated using a K ATP channel blocker. This treatment indeed reverted the protection conferred by both L-Lactate and the direct pharmacological activation of purinergic receptors (Jourdain et al., 2016;Jourdain et al., 2018), thereby implying the opening of K ATP channels via a L-Lactate/P2Y2/PI3K pathway.
These studies indicated that L-Lactate can modulate K ATP channel closure or opening depending on the context. On the one hand, L-lactate may enhance neuronal excitability and activity via K ATP channel closure (Sada et al., 2015;Karagiannis et al., 2021), whereas it promotes neuronal survival via K ATP channel opening (Jourdain et al., 2016;Jourdain et al., 2018). Although both effects modulate K ATP channels in opposite direction, they share some common features. Indeed, both enhancement of spiking activity and neuroprotection are mimicked by pyruvate (Table 1), which is also an intermediate in their signalling cascade. While the contribution of MCTs and LDH has not been evaluated on L-Lactate-induced neuroprotection per se (Jourdain et al., 2016;Jourdain et al., 2018), the observation that mitochondrial pyruvate carrier was mandatory strongly suggests that MCTs and LDH are indeed involved in neuroprotection. The mechanisms underlying both L-lactate effects start to diverge downstream of pyruvate: spiking enhancement involves an intracellular modulation of K ATP channels whereas neuroprotection entails a K ATP channel autocrine regulation.
But how could these signalling pathways deviate? A possible explanation could arise from the experimental designs leading to dramatic differences in intracellular Ca 2+ load and/or cell swelling. Indeed, the ATP release by pannexins is both Ca 2+ -sensitive and mechanosensitive (Bao et al., 2004;Whyte-Fagundes and Zoidl, 2018). In neuronal cultures, the rapid and strong intracellular Ca 2+ increase induced by excitotoxic challenges was followed by cellular swelling (Jourdain et al., 2016;Jourdain et al., 2018) and may have promoted pyruvate-derived ATP release via pannexins as well as its autocrine action on purinergic receptor. By contrast, in hippocampal slices, the effect of both glucose depletion and LDH inhibition on resting neurons was analyzed and resulted in a hyperpolarisation through K ATP channel opening (Sada et al., 2015). Under these conditions, both intracellular Ca 2+ and cell swelling are expected to be minimal (Andrew and MacVicar, 1994). Thus, the intracellular inhibition of K ATP channels induced by L-lactate or pyruvate should have been promoted. In spiking neocortical neurons (Karagiannis et al., 2021), intracellular Ca 2+ is increased by activating voltage-gated Ca 2+ channels. However, neuronal swelling is likely to be modest since increased neuronal volume is mostly observed under extreme conditions such as cordial spreading depression (Zhou et al., 2010). Altogether, these observations suggest that autocrine regulation by pyruvate-derived ATP may be favored over intracellular inhibition of K ATP channels under substantial increases in both intracellular Ca 2+ and cell volume.
Since enhancement and/or promotion of spiking activity by L-lactate is mediated by K ATP channel closure, in order to be effective this effect implicitly requires that some K ATP channels are open. This has been achieved through metabolic impairment or evoked firing (Sada et al., 2015;Karagiannis et al., 2021). Indeed, under resting conditions the neuronal ATP/ADP ratio is far above 10 and K ATP channels are already closed (Tantama et al., 2013). Metabolically-induced ATP depletion is a well established means to activate K ATP channels (Liss et al., 1999;Zawar and Neumcke, 2000;Sada et al., 2015;Karagiannis et al., 2021). However it remains unclear how continuous moderate neuronal firing (i. e., 2-4 Hz) affects K ATP channels. Indeed, in hippocampal slices brief pulses of action potentials at a frequency of up to 20 Hz resulted in an increase in P o of neuronal K ATP channels in a Na + /K + ATPase-dependent manner, which was correlated with a drop in cytosolic ATP (Tanner et al., 2011;Toloe et al., 2014). By contrast, a decrease in intracellular ATP of layer 5 pyramidal cells was only observed in vivo at stimulation frequency of 100 Hz and above (Baeza-Lehnert et al., 2019). Whether these differences relate to the preparations, the cortical areas and/or the difficulty of measuring submembrane ATP remains to be solved.
Whether ATP is the intracellular mediator of L-Lactate-inducing increase in spiking activity is also an open question. In contrast to oxaloacetate, intracellular ATP was found to be ineffective for reverting K ATP channel opening induced by LDH inhibition (Sada et al., 2015). Interestingly, hippocampal interneurons are insensitive to glucose deprivation in whole-cell configuration (Sada et al., 2015) but not in perforated patch configuration (Zawar and Neumcke, 2000), whereas the almost opposite was observed in CA1 pyramidal cells. Whether altered intracellular metabolism by whole-cell recording accounted for the apparent lack of ATP sensitivity also remains to be determined.
In summary, L-Lactate can modulate neuronal activity and excitability via different mechanisms controlling K ATP channels which depend on the metabolic context, the cell volume as well as the level of neuronal activity. In a recent study, L-lactate was also found to promote neuronal excitability in the dorsomedial prefrontal cortex via MCT2 and inhibition of large conductance Ca 2+ -activated K + channels (Yao et al., 2023). Whether both mechanisms could coexist or depend on cortical areas and developmental stages is another open question. Interestingly, in human cerebral cortex, MCT2, Kir6.2 and SUR1 expression patterns are tightly correlated, and all enriched in caudal sensory areas (Medel et al., 2022).

Integrated effect of L-lactate on neurophysiology
In the previous section, we have described different molecular mechanisms by which L-lactate modulates neuronal excitability and activity (Table 1). In the next subsections we will discuss how these different mechanisms could contribute to brain functions, namely NVC and NMC, as well as learning and memory.

NVC and NMC
The tight coupling between neuronal activity and local cerebral blood flow has been recognized more than a century ago (Roy and Sherrington, 1890). Since then, this neurovascular coupling (NVC) has been the subject of intense debates and two main contradictory hypotheses emerged. While the "metabolic" hypothesis postulated that NVC was a feedback response to an increased energy demand by neuronal activity, the "neurogenic" hypothesis attributed the neurovascular response to a feedforward process involving the release of specific vasoactive messengers (Attwell et al., 2010). The observations that: i) baseline blood flow was adequate to meet oxygen needs, even during increased neuronal activity (Mintun et al., 2001;Zhang et al., 2019;Drew, 2022), and ii) NVC persisted in an excess of blood oxygen or glucose (Wolf et al., 1997;Lindauer et al., 2010) indicated that NVC was not a response to an increase of energy demand. As a consequence, the "metabolic" hypothesis has been progressively disregarded. Nonetheless astrocyte metabolism is essential for the full process of NVC (Lecrux et al., 2011).
Neuronal activity also stimulates blood glucose uptake and its use by brain cells, a process termed NMC (for neurometabolic coupling), which relies on the interplay between glutamatergic neurons, astrocytes and blood vessels (Bonvento et al., 2002;Bonvento and Bolanos, 2021). In the late 1980s, seminal positron emission tomography studies revealed that the local increase in cerebral blood flow and brain glucose uptake induced by neuronal activity far exceeded that of oxygen consumption (Fox and Raichle, 1986;Fox et al., 1988), questioning the functional relevance of increased oxygen delivery. Thus, NVC leads to an increase in blood oxygen availability (Buzsaki et al., 2007) in the activated area, which is the physiological basis of blood oxygen-level-dependent functional magnetic resonance imaging (BOLD fMRI) (Ogawa et al., 1990) to map neuronal activity (Logothetis, 2008).
A recent study has examined the role of brain L-lactate in NMC and NVC in the rat barrel cortex (Roumes et al., 2021b). It was found that down-regulation of either neuronal MCT2 or astroglial MCT4 abolished the L-lactate increase induced by whisker stimulation. The impairment of L-lactate surge is more easily interpretable in animals presenting a down-regulation of MCT4 than in those presenting a down-regulation of MCT2. Indeed, as mentioned earlier MCT4 is selectively expressed in astrocytes (Rafiki et al., 2003;Pellerin et al., 2005;Rosafio et al., 2016) and participates in L-lactate release (Fig. 1, point 4). It is thus not surprising that altering a pathway of L-lactate release impaired the ability of astrocytes to release it, although other release mechanisms Karagiannis et al., 2016) remained unaltered. The impact of MCT2 down-regulation is however more intriguing since neuronal lactate release is expected to be very limited. An explanation might be that the uptake of L-lactate by neurons promotes the release of L-lactate by astrocytes. We indeed described before that L-lactate enhances neuronal activity via K ATP channel closure (Fig. 1, point 10), and in particular, in glutamatergic pyramidal cells (Sada et al., 2015;Karagiannis et al., 2021). Thus, entry of L-lactate in neurons through MCT2 could favor glutamate release. Yet, glutamate via its uptake in astrocytes by EAAT1/2, stimulates glucose uptake, glycolysis and L-lactate release (Pellerin and Magistretti, 1994;Voutsinos-Porche et al., 2003) (Fig. 1,  points 1-4). Thus, this would imply that feedback mechanism(s), which remain(s) to be elucidated, should exist to prevent an overproduction of L-lactate and a cortical hyperactivity.
In the same work (Roumes et al., 2021b), BOLD-fMRI-imaged NVC was also dramatically altered in rats down-regulated for MCT2, and to a lesser extent for MCT4. Impairment of the BOLD response was observed in about half of the animals with a down-regulation of MCT4. NVC could be rescued by peripheral L-lactate infusion. Conversely, the downregulation of MCT2 induced an impairment of NVC resistant to systemic L-lactate in all rats (Roumes et al., 2021b). These observations question how MCT2 and MCT4 could also be involved in NVC. Spiking activity is essential for this process by allowing the release of vasoactive messengers directly by neurons or indirectly by astrocytes and endothelial cells (Cauli and Hamel, 2010;Attwell et al., 2010;Iadecola, 2017;Cauli and Hamel, 2018;Kaplan et al., 2020). Here again neuronal lactate sensing would have promoted the release of vasoactive messengers by enhancing spiking activity (Fig. 1, point 10). This idea is supported by the rescuing of NVC by systemic L-lactate in animals downregulated for MCT4 (Roumes et al., 2021b). In the latter, lactate enters the brain most likely by endothelial MCT1 (Bergersen et al., 2002;Pierre and Pellerin, 2005;Chiry et al., 2006;Zlokovic, 2008). Along this line, systemic Llactate was found to increase motor cortex excitability in humans (Coco et al., 2010), and possibly NVC as well, which could have been mediated by neuronal lactate sensing. L-lactate could also participate in NVC by potentiating NMDA receptors (Fig. 1, point 8 and Table 1) (Yang et al., 2014;Jourdain et al., 2018). Indeed, these receptors respectively promote the release of PGE2 (Fig. 1, point 8) by cyclooxygenase-2 pyramidal cells (Niwa et al., 2000;Lecrux et al., 2011;Lacroix et al., 2015) and nitric oxide (NO) and by NO-producing interneurons (Cauli et al., 2004;Mishra et al., 2016;Lee et al., 2020), two key vasoactive messengers of NVC (Cauli and Hamel, 2010;Attwell et al., 2010;Cauli and Hamel, 2018). Furthermore, lactate-sensing and the enhancement of Llactate mediated NMDA receptors may act synergistically. By depolarizing neurons, lactate-sensing could indeed promote the removal of the Mg 2+ block from NMDA receptors (Nowak et al., 1984). In addition to the promoting effect of L-lactate on PGE2 release via K ATP channel and NMDA receptor regulation, L-lactate might also favor PGE2 accumulation by impairing its reuptake via PGT (Fig. 1, point 9), and thus enhance NVC (Gordon et al., 2008). Interestingly, systemic L-lactate potentiated the BOLD response in humans (Mintun et al., 2004), whereas the opposite was observed with systemic pyruvate (Vlassenko et al., 2006). These opposite effects may have been mediated via redox-dependent modulation of NMDA receptors by L-lactate and pyruvate (Yang et al., 2014;Jourdain et al., 2018) which could have been potentiated in the case of L-Lactate by lactate-sensing (Sada et al., 2015;Karagiannis et al., 2021) (Table 1).
In summary, L-lactate could fine-tune NVC in metabolic (i.e. K ATP ) and redox (i.e. NMDA) dependent manners ( Fig. 1 and Table 1), and may somehow reconcile the "neurogenic" and "metabolic" hypotheses of NVC. Although useful to infer neuronal activity with BOLD fMRI (Logothetis, 2008), the functional significance of NVC has been questioned (Moore and Cao, 2008;Drew, 2022). If NVC is not a response to the metabolic needs of active neurons, what other function(s) might it serve? A possible answer to this question could arise from a study showing in the rat barrel cortex that BOLD response induced by whisker stimulation was associated with the expression of c-Fos in neurons (Lu et al., 2004), an immediate early gene (IEG) whose expression correlates with neuronal activity but equally during synaptic plasticity and memory processing (Alberini, 2009). Interestingly, extracellular brain Llactate is not only doubled following neuronal activity (Prichard et al., 1991;Hu and Wilson, 1997a) but also during training of long-term memory (Suzuki et al., 2011).

Learning and memory: could L-lactate promote IEGs expression, synaptic plasticity and learning and memory?
A growing number of evidence has shown that astrocyte-derived Llactate induces the expression of plasticity-related IEGs, and potentiates glutamatergic transmission, memory formation and/or retention (Suzuki et al., 2011;Netzahualcoyotzi and Pellerin, 2020;Zhou et al., 2021;Roumes et al., 2021b;Dembitskaya et al., 2022), identifying Llactate as an important player of synaptic plasticity and memory. However, when uncoupled to memory formation L-lactate impaired memory retention (Zhou et al., 2021), thereby stressing the concept that its promoting effect on memory is time-and context-dependent (Suzuki et al., 2011;Zhou et al., 2021;Dembitskaya et al., 2022).
In both cultures of cortical neurons and the sensory-motor cortex of anesthetized adult mice, L-lactate induced the expression of the IEGs c-Fos, Zif268 and Arc (Yang et al., 2014). Similarly, glycogenolysis and Llactate release induced by optogenetically-led cAMP increase in hippocampal astrocytes have promoted the neuronal expression of Arc and c-Fos (Zhou et al., 2021). Conversely, an impairment of glycogenolysis and/or ANLS, dramatically reduced the induction of Arc or c-Fos (Suzuki et al., 2011;Zhou et al., 2021). The in vitro and in vivo observations that similar concentrations of D-lactate, pyruvate or an isoenergetic concentration of glucose was inefficient at inducing IEGs expression (Yang et al., 2014) indicated that L-lactate effect was redox-sensitive and did not rely on HCAR1 or energy metabolism (Table 1). Consistently, Dlactate impaired memory retention (Suzuki et al., 2011). However, the enhancing effects of glycogen-derived L-lactate on synaptic plasticity and memory was only found to be critical for intense neuronal activity pattern and high-attention cognitive tasks, whereas less demanding forms of synaptic plasticity and memory could rely on glucose (Dembitskaya et al., 2022).
The fact that the transcriptional regulation by L-Lactate was i) dependent on NMDA receptors and its downstream Erk1/2 (extracellular signal-regulated kinase 1/2) signaling cascade and ii) mimicked by NADH (Yang et al., 2014), showed that NMDA receptors were the primary target (Yang et al., 2014;Jourdain et al., 2018). Along this line, the longterm potentiation underlying the promoting effect of L-lactate on memory relied on NMDA receptors (Suzuki et al., 2011;Zhou et al., 2021;Dembitskaya et al., 2022). Furthermore, the observation that the impairment of long term potentiation by intracellular LDH inhibition was rescued by NADH but not by pyruvate, demonstrated that the effect of L-lactate was dependent on redox state and not on downstream pyruvate metabolism (Dembitskaya et al., 2022). Altogether, these findings indicated that the promoting effect of L-lactate on synaptic plasticity was primarily meditated by its action on NMDA receptors (Table 1). Nonetheless, as discussed earlier on NVC, the depolarization induced by K ATP channel closure could have contributed to the effect of L-lactate by favoring the removal of the Mg 2+ block of NMDA receptors (Nowak et al., 1984). Interestingly, the effect of L-lactate on synaptic plasticity was observed with high activity patterns (Dembitskaya et al., 2022), which notably induce intracellular ATP drops (Baeza-Lehnert et al., 2019) thereby favoring the opening of K ATP channels (Tantama et al., 2013) which is essential for lactate-sensing (Sada et al., 2015;Karagiannis et al., 2021). Although, the effect of L-lactate on long term potentiation does not depend on pyruvate metabolism on (Dembitskaya et al., 2022), the use of whole-cell patch-clamp may have altered intracellular metabolism and render modulation of K ATP channels ineffective.
In addition to inducing the expression of IEGs, L-lactate was also found to control adult hippocampal neurogenesis (Scandella and Knobloch, 2019;Nicola and Okun, 2021) that is key to the formation of spatial memories (Netzahualcoyotzi and Pellerin, 2020). Both adult hippocampal neurogenesis and memory formation involved the neuronal uptake of L-lactate via MCT2, but not HCAR1 (Lev-Vachnish et al., 2019;Netzahualcoyotzi and Pellerin, 2020). Interestingly, Llactate was also found to induce the expression of Brain Derived Neurotrophic Factor (BDNF) in neuronal cultures (Yang et al., 2014) : BDNF expression is importantly induced in the late transcriptional phase of plasticity processing (Caroni et al., 2012) as well as in neurogenesis (Lev-Vachnish et al., 2019). The beneficial effect of physical exercise on learning and memory was also found to be mediated by blood-borne lactate and BDNF induction (El Hayek et al., 2019). It is worth noting that the production of BDNF is altered in Kir6.2 knockout mice and impaired by a K ATP channel opener (Fan et al., 2016), both conditions compromising lactate-sensing. Furthermore, Kir6.2 knockout mice have an altered response to novelty (Deacon et al., 2006), which is at the basis of memories promoted by L-lactate (Zhou et al., 2021;Roumes et al., 2021b;Dembitskaya et al., 2022). Determining whether K ATP channels indeed contribute to the promoting effect of L-lactate on memory would require further investigations.

Deregulation of L-lactate signaling in pathologies
In this section we will briefly report on how different mechanisms regulating neuronal excitability and activity by L-lactate are deregulated in some neurological diseases. We will first focus on two pathologies, epilepsy and stroke, in which neuronal, astroglial, metabolic and vascular deficits overlap, thereby altering the integrated functions modulated by L-lactate described in the previous sections. We will also discuss how the different mechanisms of L-lactate signaling could cooperate or oppose in pathogenesis. Finally, we will list some examples of other pathologies in which lactacte dysregulation is involved.

Epilepsy
Brain L-lactate is dramatically increased in a rat model of status epilepticus (Slais et al., 2008). K ATP channels and lactate-sensing were indeed found to be critically involved in epilepsy, and stiripentol, an antiepileptic drug, was revealed to be a LDH inhibitor (Sada et al., 2015). In addition, HCAR1 could also be involved in seizures (Briquet et al., 2022), although in an opposite manner compared to K ATP channels. Whether both mechanisms co-exist and compete remains to be determined. In addition, a deleterious postictal hypoperfusion, which is used to infer the localization of seizure onset with perfusion MRI (Gaxiola-Valdez et al., 2017), was found to critically involve PGE2 derived from cyclooxygenase-2 activity and the activation of its EP1 vasoconstrictive receptor (Tran et al., 2020). Along this line plasma levels of PGE2 are elevated in epileptic patients (Rawat et al., 2020) suggesting an increased brain PGE2 synthesis. Increased parenchymal PGE2 concentration may indeed favor vasoconstriction over vasodilation by progressively recruiting the arteriolar vasoconstrictive receptor EP1 (Dabertrand et al., 2013;Czigler et al., 2020;Rosehart et al., 2021) that shows lower affinity than the EP2 and EP4 dilatory receptors of PGE2 (Boie et al., 1997;Lacroix et al., 2015;Mishra et al., 2016). In addition, an excess of PGE2 can also lead to EP4 receptor internalization (Desai et al., 2000), thereby decreasing its vasodilatory potency. As described in the NVC and MNC section, L-lactate could have contributed to PGE2 overproduction and accumulation by potentiating NMDA receptors (Yang et al., 2014;Jourdain et al., 2018), by enhancing neuronal activity via K ATP channel closure (Sada et al., 2015;Karagiannis et al., 2021) and/or by impairing PGE2 clearance (Gordon et al., 2008).

Ischemia
Ischemia is another example of pathology in which the modulation of neuronal activity and excitability by L-lactate is involved. Indeed, Llactate was found to be neuroprotective against ischemia at moderate levels in adult mice and neonatal rats (Berthet et al., 2009;Roumes et al., 2021a), but toxic at a higher doses (Berthet et al., 2009). This dual effect of L-lactate presumably involves opposite mechanisms in a dosedependent manner. The beneficial effects were likely to be mediated by the activation of K ATP channels (Jourdain et al., 2016;Jourdain et al., 2018) and/or HCAR1 receptor (Bozzo et al., 2013;de Castro Abrantes et al., 2019) which might have mitigated the deleterious stroke-induced energy demand through lowering neuronal excitability and/or activity. Along this line both K ATP channels and HCAR1 receptor were found to be neuroprotective against ischemia (Sun et al., 2007;Castillo et al., 2015). Since pyruvate also conferred neuroprotection (Castillo et al., 2015), the beneficial effect of L-lactate could also be exerted by promoting ATP production, thereby counterbalancing stroke-induced energy shortage. On the other hand, the deleterious effects of L-lactate higher doses (Berthet et al., 2009) may involve lactate-sensing (Sada et al., 2015;Karagiannis et al., 2021), might aggravate the deleterious strokeinduced energy demand by enhancing neuronal activity.

Conclusion
In summary, L-lactate can modulate neuronal excitability and activity by acting on at least three main molecular mechanisms. While the activation by L-lactate of HCAR1 decreases neuronal activity, the modulation of NDMA receptors and K ATP channels by L-lactate generally promotes neuronal activity and/or excitability ( Fig. 1 and Table 1). Thus, L-lactate is able to contextually fine-tune various brain functions, depending on i) its systemic or cerebral source, the latter deriving from astrocyte glycolysis and/or glycogenolysis, ii) the relative distribution of its molecular targets across cortical areas, and iii) the level of neuronal activity. The effects of lactate rely on a delicate interplay between neurons, astrocytes, blood vessels and metabolism. The recent advances on the molecular and cellular mechanisms underlying L-lactate modulation of neuronal excitability and activity will allow to decipher how these different mechanisms are altered in several neurological diseases. Determining how the different mechanisms of L-lactate signaling are altered in these pathologies could open new promising avenues for therapeutic targets.

Funding
Bruno Cauli is supported by the i-Bio collaborative grant of Sorbonne University. Dongdong Li is supported by a grant from the Agence Nationale de la Recherche (ANR-20-CE14-0025). The funders had no role in the writing or the decision to submit the work for publication.

Declaration of Competing Interest
The authors declare no competing interests.

Data availability
No data was used for the research described in the article.