- Our team was the first one to show that KCC2 is the major chloride transporter in CNS neurons, accounting for the generation of hyperpolarizing IPSPs, as well as their emergence during brain development (Rivera et al., 1999). This study launched whole new areas of research at the molecular, cellular and whole-organism levels with important implications on neuronal development, plasticity and disease (reviews Kaila et al., 2014 a, b). The developmental shift of GABA action from depolarizing to hyperpolarizing has been recently demonstrated in vivo by means of two-photon imaging of intracellular chloride in pyramidal neurons (Sulis Sato et al., 2018).

- We were the first to demonstrate that GABAergic currents under physiological conditions are mediated by both chloride and bicarbonate ions (Kaila and Voipio, 1987; review Kaila, 1994). This radically modified the classical static view of synaptic inhibition entrenched by the Nobel Laureate, Sir John Eccles. HCO3-carries a substantial inward (depolarizing) current across GABAARs because the equilibrium potential of bicarbonate (EHCO3) is set by neuronal pH-regulatory mechanisms to a rather positive level, around -15 mV, making EGABA more positive than ECl (Kaila and Voipio, 1987; Kaila et al., 1989; Kaila et al., 1993). This, in turn, leads to depolarizing GABAergic responses in mature KCC2-expressing neurons that have a high resting potential, such as neocortical neurons in slice preparations (Kaila et al., 1993). (Unfortunately, the above mechanism has sometimes beenconfused with the more complex hyperpolarizing-to-depolarizing mode of GABA action described below.)    

- Moreover, GABAergic excitation in mature neurons with a biphasic hyperpolarizing-to-depolarizing response involves an even more elaborate cascade of events (Kaila et al., 1997; Viitanen et al., 2010): The efflux of HCO3- is promoted by the action of neuronal carbonic anhydrase (first identified by us in pyramidal neurons; Pasternack et al., 1993) which further facilitates theuptake of Cl-and leads to an increase in extracellular K+. This makes EGABAprogressively more depolarizing, followed by a direct depolarizing action of the extracellular K+.  

- The abrupt developmental expression of the neuron specific carbonic anhydrase 7 (CA7) turned out to be a crucial factor in setting the time of enhanced susceptibility towards epileptiform activity in rodents at around postnatal day 10-12 which, in terms of cortical development corresponds to the human newborn, where seizures are a major and unmet clinical problem. CA7 was shown to be crucial in the developmental emergence of the GABA-mediated pro-excitatory extracellular potassium transients (Ruusuvuori et al., 2004). Notably, potassium transients have been known for decades to be a hallmark of epileptic seizures and thus, our work points to a key role for GABAergic excitation in the generation of seizures. 

- The original findings above provided an explanation (there are probably several mechanisms) for the frequent refractoriness of seizures to classical GABA-enhancing anti-epileptic agents (Puskarjov et al., 2014b). Another, non-exclusive explanation for pharmacoresistance is the down-regulation of the neuron-specific K-Cl cotransporter KCC2 (Rivera et al., 2004; Puskarjov et al., 2012) in brain tissue following seizures and other types of pathophysiological states, such as stroke.

The timing of the developmental upregulation of KCC2 with regard to birth is both brain region- and species-specific. Rats and mice are born with very low KCC2 expression and depolarizing GABAAR-mediated responses in cortical neurons, while in the guinea pig, KCC2 is upregulated in utero and cortical neurons show hyper­polarizing GABAAR responses at birth (Rivera et al., 1999; Kaila et al., 2014a). In the human neocortex, SLC12A5mRNA which codes for KCC2 protein undergoes robust upregu­lation during the second half of gestation, and there are immunohistochemical data indicating that from the 25th postconceptional week onwards, most cor­tical neurons express KCC2. Thus, unlike in rodents, massive upregulation of KCC2 in the human neocortex begins prenatally (Sedmak et al., 2016). This finding is in agreement with the fact that the shift from discontinuous to continuous cortical EEG occurs in humans around full-term birth (Vanhatalo et al., 2002; Vanhatalo et al., 2005) and in rats at about P11-P12. Moreover, our recent work, shows that functional KCC2 is expressed at a low but physiologically significant level in P0 rats and mice, with a marked effect in modulating spontaneous network events and thus formation of the cortex. In terms of human cortical development, rat and mouse P0 corresponds to postconceptional weeks 25-27.       

- KCC2 is a very useful indicator of the stage of maturation of a neuron within a given population, and also between distinct neuronal types (Rivera et al., 1999). In addition, down-regulation of KCC2 expression seems to faithfully reflect the neuronal dedifferentiation which takes place in various diseases and trauma (Kaila et al., 2014a). Whether this down-regulation, which leads to depolarizing GABA responses, is adaptive or maladaptive is not known (Kaila et al., 2014b). However, the widely spread idea that KCC2 down-regulation is disease-promoting has little empirical support and raises questions regarding general evolutionary and developmental mechanism that promote adaptation, such as minimizing neuronal energy consumption and maximizing neuronal survival in response to an insult (Buzsaki et al., 2007).

- We were the first to describe a human disease mutation of KCC2, which results in impairments in both GABAergic and glutamatergic signaling and is a susceptibility variant related to febrile seizures and generalized idiopathic epilepsy (Puskarjov et al., 2014). The mutation affects both the ion-transport function and the morphogenic role of KCC2.  This work has an important evolutionary dimension provided by the fact that the “genic intolerance” for variability of KCC2 is among the lowest ones in the human genome.

- Unlike KCC2, the Na-K-2Cl cotransporter NKCC1 is expressed in most tissues and cells. In most neurons, it is the main transporter that accumulates Cl (Kaila et al., 2014a). The available data on the developmental regulation of NKCC1 are highly contradictory. We are now using cell-type specific detection of NKCC1 and its isoforms (NKCC1a and b) to analyze their spatiotemporal expression patterns during brain development.

- Both KCC2 and NKCC1 are involved in myriads of neuronal responses and network events. For instance, GABAergic depolarization and excitation in the neonate hippocampus drives, together with glutamatergic pacemakers, the so called Giant Depolarizing potentials (GDPs), which were initially first described by Enrico Cherubini and Yehezkel Ben-Ari in 1989. Notably, these events (and their more mature counterparts, the sharp positive waves, SPWs) are blocked by the neurohormone arginine vasopressin (AVP), which led us to hypothesize that fetal-endogenous AVP has a pre-emptive neuroprotective effect in the perinatal mammalian during parturition (Spoljaric et al., 2017; and see below). In adult neurons, NKCC1-dependent depolarizing (but not necessarily functionally excitatory) actions take place in the axon initial segment (Khirug et al., 2008).  

- Importantly, some neuronal ion-regulatory proteins (IRPs) are not only responsible for Cl-and pH homeostasis, but we have obtained evidence that two neuron-specific IRPs (KCC2 and CA7; see also below) act also as structural proteins that directly sculpt neuronal morphology. In collaboration with Claudio Rivera we initially showed that KCC2 has an ion-transport independent, structural role in the formation of dendritic spines in cultured cortical neurons (Li et al., 2007; see also Horn et al., 2010). This observation was later verified under in vivo conditions, and it turned out that only a part of the large cytosolic C-term of KCC2 was sufficient to promote spinogenesis (Fiumelli et al., 2012). The dual roles of KCC2 in the ontogeny of hyperpolarizing IPSPs and in dendritic spinogenesis suggest that KCC2 acts as a coordinating factor in the development of functionally inhibitory and excitatory transmission. Notably, a reversal of this process (shift from hyperpolarizing to depolarizing GABAergic inhibition and loss of dendritic spines) has been reported in pathophysiological conditions (Kaila et al., 2014a; b).

- The cytosolic carbonic anhydrase isoform CA7 was first identified as a neuronal isoform by our lab and was suggested to play a major role in the development of HCO3- -dependent GABAergic depolarization (Kaila et al., 1997; Ruusuvuori et al., 2004; Viitanen et al., 2010). Using a novel CA7 KO, a CA2 KO and a novel CA2/7 double KO mouse, we found that neuronal CA activity is due to two cytosolic isoforms, CA7 and CA2 (Ruusuvuori et al., 2013).  CA2 and CA7 show striking developmental and cell-biological differences: the neuron-specific CA7 appears to have a dual role in the modulation of pHi and of actin cytoskeleton. We are currently examining the subcellular localization of CA7 as well as the actin interaction of novel CA7 variants generated in the laboratory. Furthermore, using the recently described CA7 deficient mice (Ruusuvuori et al., 2013) we are able to specifically study the effects of CA7 on dendritic spine morphology and dynamics, plasticity and excitability of neurons and neuronal networks, and on seizure susceptibility.

- We and many others have found that intra- and extracellular pH changes within brain tissue can have an extremely potent action on neuronal and network excitability. Alkalosis enhances while acidosis suppresses excitability by affecting a large spectrum of molecular targets, including voltage- and transmitter-gated channels (Kaila and Ransom, 1998; Chesler and Kaila, 1992; Ruusuvuori and Kaila, 2014). The fact that a number of physiological factors can result to global (breathing, energy metabolism) as well as highly localized changes (synaptic transmission, spiking) highlight the importance of proton modulation of neuronal signaling (Ruusuvuori and Kaila, 2014) in health and disease.We have found that an intraneuronal change of 0.05 pH units is sufficient to have a profound effect on spontaneous network activity in neonatal hippocampal slices (Ruusuvuori et al., 2010). pH is a very powerful modulator of brain excitability and its role in epilepsy has been recognized for more than a century. 

Febrile seizures

- Respiratory alkalosis is known to precipitate seizures, especially in children. Using an established model of experimental febrile seizures (FS) based on hyperthermia-exposed immature rats, we showed that FS were associated with a respiratory alkalosis caused by hyperventilation in response to the elevated temperature (Schuchmann et al., 2006). A cause-effect between the alkalosis and seizure generation was indicated by the fact that a brief (20 second) exposure to 5% CO2 in air completely blocked both the behavioral and EEG seizures. More recently (Ruusuvuori et al., 2013), we have shown that juvenile mice lacking the neuronal CA7 isoform do not have electrographic FS at all while clinical seizures are only mildly reduced – a finding that raises questions regarding the origin of the clinical seizures (Pospelov et al., 2016).  

- The clinical relevance of the above work is supported by our retrospective study on age-, gender- and fever-matched children admitted to the hospital because of either FS or gastroenteritis (GE) (Schuchmann et al., 2011). While blood acid-base parameters showed a respiratory alkalosis 30 min after FS, children with GE were acidotic. In accordance with alkalosis as a cause of FS, none of the children with GE had FS. Moreover, children with a known susceptibility to FS did not have FS if their fever was caused by GE, which further indicated that the GE-induced acidosis inhibits the occurrence of FS. Thus, a simple putative therapy for acute treatment of FS might be based on elevation of respiratory CO2 (Schuchmann et al., 2006).

- During birth in mammals, a pronounced surge of stress hormones in the blood takes place to promote survival in the transition to the extrauterine environment. However, it is not known whether signaling by AVP at birth also involves central pathways with direct protective effects on the brain. We showed recently (Spoljaric et al., 2017) that AVP specifically activates interneurons to suppress spontaneous network events (GDPs in the rat and SPWs in the guinea pig; see section 2) in the perinatal hippocampus.  The effect of vasopressin is not dependent on the level of maturation (depolarizing vs. hyperpolarizing) of postsynaptic GABAA receptor responses. Instead, AVP activates interneurons to desynchronize both types of CA3-driven events. This seems to be an evolutionarily conserved mechanism that is well-suited to suppress energetically expensive correlated network events under conditions of reduced oxygen supply at birth.

- Birth asphyxia, estimated to account for a million neonatal deaths annually, can cause a wide variety of neurodevelopmental impairments resulting in neuropsychiatric disorders and diseases. Thus, there is a need to develop new methods to swiftly identify those neonates who would benefit from neuroprotective treatments such as hypothermia, which are intended to alleviate hypoxic-ischemic encephalopathy (HIE). AVP secretion is steeply enhanced during the “obligatory period” of asphyxia which takes place during the shift from placental to lung-based breathing in all mammals. In protracted and/or complicated birth, a clinical condition emerges which is (also) termed – and diagnosed – as “birth asphyxia” but should not be confused with the obligatory asphyxia mentioned above. Under the more severe conditions, AVP secretion is further enhanced, which can be monitored using a chemically stable fragment of prepro-AVP, copeptin (Fig). In our collaborations with clinicians in Helsinki, we have found that copeptin has the potential of becoming a diagnostic marker of clinical birth asphyxia (Summanen et al., 2017).  In a subsequent collaboration with the team of Miklos Szabo in Budapest we found that copeptin holds much promise as a prognostic biomarker, as seen in the inverse correlation between umbilical blood copeptin levels and the Bayley Scales of Infant and Toddler Development-II at 2 years of age (Kelen et al., 2017).

- We have also shown that normal birth in rats is accompanied by a surge in copeptin levels, which is massively enhanced by experimental asphyxia (Summanen et al., 2018). These results provide further mechanistic insights on birth- associated endocrine responses, and they further validate our translational rodent models.   

- We have very recently designed a novel, non-invasive model of birth asphyxia based on exposure of postnatal (P) day 11 rats to asphyxia-like conditions, brought about by an increase of CO2 to 20% and a decrease of O2 to 5% in ambient air. This model is more severe than the one we published before (Helmy et al., 2011) and better mimics the conditions seen in human neonates diagnosed with birth asphyxia and hypoxic-ischemic encephalopathy (HIE). 

-  Immediately after the termination of the 30 min exposure to this gas mixture, pronounced behavioral tonic-clonic seizure activity is seen in parallel with cortical electrographic seizures, which are strictly related to the alkaline-directed recovery from the asphyxia-induced acidosis. Strikingly, slowing down the recovery of the post-asphyxia by applying exogenous CO2 completely abolishes the seizures. This suggests a novel resuscitation strategy based on Graded Restoration of Normocapnia (GRN) (see Helmy et al., 2011) following birth asphyxia. We are using both P11 and P6-7 rats in these studies to be able to examine the qualitative differences of asphyxia effects at developmental stages corresponding to preterm and full term human neonates.   

- Using extracellular microelectrodes for measurement of brain and trunk (subcutaneous or “body”) pH as well as blood gas analyses (see Helmy et al., 2011), we have done extensive studies (unpublished) on acid-base and O2 shifts in our P11 rat asphyxia model. Under conditions designed to mimic an abrupt onset of asphyxia, brain pH shows a very rapid fall which recovers promptly upon re-establishment of control conditions. “Body pH” falls more slowly and has a prolonged component that most likely reflects lactate production as seen also in the levels of the negative base excess in blood. Our data demonstrate a compartmentalization of brain and body pH at the level of the blood-brain barrier (BBB). Interestingly, and in line with data obtained in adult humans (Voipio et al., 2003), the DC-potential across the BBB reflects the time course of the above pH shifts.

- During the experimental asphyxia, the O2 levels plummet to levels too low to be even detected in the brain. Notably, the oxygenation of brain tissue is strongly enhanced by CO2 during both recovery from asphyxia and control conditions, which shows (among other things) that birth asphyxia cannot be mimicked by experimental paradigms based on hypoxia only. 

- In clinical practice, the artifactual DC shifts that tend to contaminate the recordings are eliminated simply by high-pass filtering of the recorded signals. Unfortunately, this also means that lots of relevant neurophysiological and pathophysiological information is completely lost. Our team has been actively developing DC-EEG recording methods (Voipio et al., 2003) for routine use in the hospital, especially for characterization of preterm and neonatal EEG (Vanhatalo and Kaila, 2006), where detection of EEG abnormalities and seizures still is a major challenge.

- DC-EEG recordings also showed that gross excitability and interictal events in the human cortex are modulated by infraslow oscillations (Vanhatalo et al., 2004).

- Large DC-EEG signals are not of neuronal origin, and the blood-brain barrier (BBB) is able to generate millivolt-range DC-shifts on the scalp during for instance hyperventilation (Voipio et al., 2003). These signals are attributable to pH shifts at the BBB, and we are currently examining their utility for indirectly monitoring the brain pH changes that take place during experimental asphyxia. We also aim at developing a simple device that could be used in the delivery room and/or NICU to obtain a DC-EEG read-out of severity and time course of pH changes related to clinical birth asphyxia and HIE.