Formulating treatment of major psychiatric disorders: algorithm targets the dominantly affected brain cell-types

Background Pharmacotherapy for most psychiatric conditions was developed from serendipitous observations of benefit from drugs prescribed for different reasons. An algorithmic approach to formulating pharmacotherapy is proposed, based upon which combination of changed activities by brain cell-types is dominant for any particular condition, because those cell-types contain and surrogate for genetic, metabolic and environmental information, that has affected their function. The algorithm performs because functions of some or all the affected cell-types benefit from several available drugs: clemastine, dantrolene, erythropoietin, fingolimod, fluoxetine, lithium, memantine, minocycline, pioglitazone, piracetam, and riluzole Procedures/findings Bipolar disorder, major depressive disorder, schizophrenia, Alzheimer’s disease, and post-traumatic stress disorder, illustrate the algorithm; for them, literature reviews show that no single combination of altered cell-types accounts for all cases; but they identify, for each condition, which combination occurs most frequently, i.e., dominates, as compared with other possible combinations. Knowing the dominant combination of altered cell-types in a particular condition, permits formulation of therapy with combinations of drugs taken from the above list. The percentage of patients who might benefit from that therapy, depends upon the frequency with which the dominant combination occurs in patients with that particular condition. Conclusions Knowing the dominant combination of changed cell types in psychiatric conditions, permits an algorithmically formulated, rationally-based treatment. Different studies of the same condition often produce discrepant results; all might be correct, because identical clinical phenotypes result from different combinations of impaired cell-types, thus producing different results. Clinical trials would validate both the proposed concept and choice of drugs.


Introduction
It is an old debate as to whether or not psychiatry is a branch of medicine or should be regarded as totally separate; also debated are queries concerning confidence in both diagnostic systems and treatment strategies [1,2]. This article is not intended as a polemic in those debates but is aimed at a rational formulation of pharmacotherapy for patients whose cases receive only limited benefit from psychotherapy. It is relevant to note that effective pharmacotherapy for major psychiatric disorders commenced after serendipitous observations that chlorpromazine, originally produced and prescribed as an antihistamine, was seen to reduce symptoms of mania and schizophrenia; that imipramine, originally

Altered functional activity of dominant cell-types in selective psychiatric disorders used as illustrations
Although the concept suggested in this article is an algorithmic approach for formulation of pharmacotherapy of any psychiatric condition, five of those are considered for illustrative purposes. Decreases in either their number or activity, affecting astrocytes, oligodendrocytes, neurons, and endothelial cells, would predispose to disease, whereas for microglia, increase in their number or functional activity would predispose to disease; changes in opposite direction would likely not be deleterious and, therefore, are not mentioned here. Note that the determination of the dominant change in the brain cell-type is based upon the findings in citations provided by PubMed and Google Scholar, supplemented by those in key articles.
Oligodendrocytes Liu et al. found no increases in any report and decreases in 2 [21], and Uranova et al. saw a 29% decrease [22].
Interneurons Gigase et al. summarized reported studies of BD brains, showing interneurons as increased in one brain region, and decreased in sixteen [23].
Endothelial cells Kruger et al. measured CBF by using [150] H 2 O and PET scanning [39]. Subjects drafted a brief script describing a sad life event; the script was projected onto a computer screen during the PET scan; CBF was measured in baseline euthymia and in provoked sadness. Decreased CBF in the ventromedial (vm) PFC distinguished BD patients from their at-risk siblings. Another study used breath-holding, which should produce decreased cerebral oxygenation and cerebral vasodilatation. Twenty-five adolescents with BD and 25 psychiatrically healthy controls completed six 15-s breath-holds during functional magnetic resonance imaging, and cerebrovascular reactivity (CVR) was determined by comparing blood-oxygenation-level dependent signal changes [40]. During breath-holding, CVR in several brain regions was decreased in BD versus controls.
A systematic review by Toma et al., yielded 33 studies, 118 subjects with BD/depression, 46 with BD/mania, 24 with BD/ euthymia, and 385 controls [41]. CBF in multiple brain areas was reduced in all BD subjects except 22 who had increased flow. However, discrepant results were found by Zeng et al., who saw no difference in CBF, as assessed by arterial spin labelling, in 61 patients with BD [42].
Microglia Liu et al. found increases in 1 report and decreases in 2 [21]. Summary: the dominant changes in brain cell-type in BD are decreases in oligodendrocytes, neurons, and endothelial cells.

AD
Astrocytes in AD. Emphasizing their importance for cognition, astrocyte numbers in the dentate gyrus were more reduced in Braak stages 3-4 than in stages 0-2 [66]; and because astrocyte processes wrap around cerebral microvasculature, the morphological modifications of astrocytes affect micro-cerebral blood flow [67] and, therefore, neural function [68].
Oligodendrocytes in AD. Mature oligodendrocytes perform myelination of naked axons; if their numbers are decreased, e.g., from impaired maturation of oligodendrocyte precursor cells (OPC), so that myelination becomes inadequate, then neural tracts suffer and cognition may be disturbed [68]. Exposure of oligodendrocyte cultures to 1 μM of Aβ 1-42 induced cell death; morphological changes, e.g., shrunken cell bodies and a breakdown of their processes; and an 3.2-fold increase of lactate dehydrogenase activity released to the culture media [69].
Neurons in AD AD patients with severe tau pathology had a decreased number of newly generated neurons in their dentate gyrus [66]. Leng et al. found a selectively vulnerable subpopulation of excitatory neurons in the entorhinal cortex (EC) and showed a depletion of this subpopulation during AD progression [70].
Endothelial cells in AD Microvascular disease affects the progression of cognitive deficits in AD [71]. Endothelial cells contain the scavenge receptor, CD36, whose interaction with Aβ activates NADPH oxidase, producing vascular oxidative stress and neurovascular dysregulation [72]. String vessels, the remnants of destroyed microcapillaries, are seen throughout the brain of AD [18]. In addition, microinfarcts, with minute foci of neuronal loss as small as 50 µm, were seen in 43% of AD brains [73].
Microglia in AD Several reports concerning the role of microglia in AD are conflicting, the likely reason being the need to distinguish between proinflammatory and anti-inflammatory microglia, because a skewed M1 activation over M2 has been related to disease progression in AD [20], and this is the dominant but not universal change in AD.
Summary: the dominant changes in brain cell-type in AD are decreases in astrocytes, oligodendrocytes, neurons and endothelial cells, and increases in microglia.

PTSD
Astrocytes: Gill et al. assessed astrocyte levels in PTSD and found them decreased [74]. Chronic stress in rats reduced the length and volume of astrocytic processes by 40.6% and 56%, respectively, and the number of their branch points was reduced by 57.8% [75].
Oligodendrocytes: Brains from females with PTSD had down-regulated oligodendrocyte-associated genes (MBP and MBOP) [76]. Bonnefil et al. showed that social defeat stress in mice reduced mature oligodendrocytes by 31.8% [77]. In another study they showed that chronic variable stress produced down-regulation of oligodendrocyte specific genes in the nucleus accumbens (NAcc) and PFC but upregulation in the corpus callosum [78]. Banasr et al. caused unpredictable stress in rats; after 15 days, the prelimbic cortex had 21% fewer oligodendrocytes [16]. After traumatic brain injury in rats, numbers of oligodendrocytes decreased significantly [79].
Neurons: Smith et al. produced fear conditioning in mice, causing reduced density of dendritic spines of pyramidal neurons in the hippocampus and medial PFC [80].
Endothelial cells: Brains from persons with PTSD had up-regulated endothelin1 gene, EDN1, cells in the vmPFC, which would decrease cerebral microvascular flow [76]. After traumatic brain injury in mice, there was a 1.5-fold increased apoptosis of cerebrovascular endothelial cells and increased BBB permeability [81].
Microglia: Fear conditioning also increased the length of microglial processes [80]. Rats whose behavior was severely disturbed by the scent of predators, had increased hippocampal microglia [82]. After single, prolonged stress, even minimal pain induced activation of hippocampal microglia [83].
Minocycline prevented oligodendrocyte toxicity caused by deprivation of oxygen and glucose [184], microglialinduced apoptosis [185], by minocycline preventing inhibition of CREB's up-regulation [186] and caused by Aβ [187]. Synaptic function improved from increased levels of PSD-95 and dendritic spines [188]. It prevented cognitive decline caused by an antagonist of the NMDAR [189]. Neuronal benefit was also from its increased expression of BDNF, CREB, and phospho-CREB [186], and proliferation of NPCs [190]. Minocycline potentiated neurite outgrowth [191]. In a transgenic mouse model of Down's syndrome, minocycline prevented decline of cholinergic neurons [192]. In a list of 1040 drugs that prevent release of cytochrome c, minocycline was the second most potent [193], the mechanism involves inhibiting mitochondrial increases in Ca 2+ concentration plus inhibition of NADH-cytochrome c reductase and cytochrome c oxidase [193,194]. Finally, minocycline prevented microglial activation [195,196].
Piracetam increased numbers and function of astrocytes [209]. It also increased function of synapses [210][211][212][213] and neurons [210]: it decreased neurotoxicity from deprivation of oxygen and glucose, hypoperfusion, ethanol feeding or ethanol withdrawal [211,[214][215][216] It also caused longer neurites [210]. The effects of piracetam may derive in part 1 3 from the restoration of cell membrane fluidity induced by a conformational change in the phospholipids of liposomal membrane [213] By decreasing mitochondrial swelling and permeability caused by excessive Ca 2+ opening the MPTP, it improved mitochondrial membrane potential and ATP levels, shifted the balance of mitochondrial fission or fusion towards fusion, and reversed the adverse effect of pro-oxidants [217]. Additionally, it reversed the cytotoxic effects of p53 and BAX [211].
Riluzole increased both the gene for the excitatory amino acid transporter (EAAT2) and its expression, thus enhancing glutamate uptake by astrocytes, and protecting against excitotoxicity of neurons [218,219]. It benefits synapses, by inhibiting voltage-activated sodium currents which prevents reverse operation of the Na + /Ca 2+ exchanger [220], and by increasing LTP [221]. It caused a 40-fold increase in hippocampal BDNF, creating neurogenesis [222]; and protected against neural degeneration caused by ischemia [223]. N-acetylaspartate is exclusively expressed in neurons and was increased in the cerebral cortex by exposure to riluzole [224]. In neurons, it also decreased oxidative stress, lipid peroxidation, and ATP depletion [225,226]. Microglial activation was ameliorated by riluzole [227], which also upregulated the mRNA levels of M2 markers and downregulated those of M1 markers [228].

Drugs that might be therapeutically beneficial in a reasonable number of patients, by addressing the dominantly affected cell-types in BD, MDD, SCZ, AD, and PTSD
For BD, increasing oligodendrocytes, neurons and endothelial cells (CBF), is achieved by erythropoietin, fluoxetine, lithium, memantine, and pioglitazone. For MDD, increasing astrocytes, oligodendrocytes, neurons, and endothelial cells (CBF), and decreasing microglia, is achieved by erythropoietin, fluoxetine, lithium and pioglitazone.
For AD, increasing astrocytes, oligodendrocytes, neurons, and endothelial cells (or CBF), and decreasing microglia, is achieved by erythropoietin, fluoxetine, lithium and pioglitazone.
For PTSD, increasing astrocytes, oligodendrocytes, neurons, and endothelial cells, and decreasing microglia, is achieved by erythropoietin, fluoxetine, lithium and pioglitazone.

Shared dominant cell-types between disorders
There are shared changes in dominant cell-types between some psychiatric conditions, from which one may infer shared pathogenesis and similar treatments. Thus, MDD, AD, and PTSD, share decreases in astrocytes, oligodendrocytes, neurons, and endothelial cells, and increases in microglia. SCZ has decreases in oligodendrocytes, neurons, and endothelial cells, and increases in microglia, so shares these features with MDD, AD, and PTSD. Evidence of an overlapping genetic risk that is similar to their overlapping of affected cell-types, was shown for MDD and AD by Lutz et al. [229], between PTSD and schizophrenia by Duncan et al. [230], and between MDD and schizophrenia by The Psychiatric Genomics Consortium [231]. That consortium demonstrated significant genetic correlations between SCZ and BD (0.68), MDD and BD (0.47), SCZ and MDD (0.43), and was non-significant between psychiatric disorders and the negative control, Crohn's disease. The foregoing genetic data support the concept proposed in this article, that brain cell-types, may be used as surrogates for all of the genetic, metabolic, and environmental, data for the inputs of which they are storehouses.

APOE4 affects functions of astrocytes, oligodendrocytes, neurons, endothelial cells, and microglia
Further support for the concept that brain cell-types may be used as surrogates for several abnormalities affecting their function, comes from brain cell-types that carry APOE4. The APOE4 variant of the APOE gene product, renders its carriers susceptible to AD. One mechanism for this may be its effect on the five families of brain cell-types. Blanchard et al. reported several such effects: that cholesterol transport was defective in oligodendrocytes derived from induced pluripotent stem cells (iPSC) carrying APOE4; that these APOE4+oligodendrocytes accumulated cholesteryl ester species; and that this was associated with down-regulation of genes linked to myelination and decreased production of MBP [232]. Lin et al. showed that the APOE4 variant can also lead to extensive gene expression alterations in neurons, astrocytes, and microglia; APOE4 astrocytes had cholesterol accumulation, and APOE4 microglia-like cells had increased up-regulation of genes associated with immune responses and inflammation [233]. Lipid accumulation in APOE4+microglia was seen also by Victor et al.; and, 1 3 as compared with that of APOE3 microglia, the pro-inflammatory activity of APOE4+microglia was exacerbated by medium from neuronal cultures, demonstrating cross-talk between neurons and APOE4+microglia [234]. APOE4+endothelial cells had dysfunctions in pathways leading to blood clotting factors and inflammation [235]. Finally, removal of APOE4 from astrocytes led to improvements in synaptic function with increase in PD95, neurons, oligodendrocytes, and microglia [236]. Thus, all the foregoing data suggest that one of the several mechanisms accounting for the association between APOE4 and AD, is that APOE4 detrimentally affects the functions of astrocytes, oligodendrocytes, synapses, neurons, endothelial cells, and microglia, which supports the thesis of this article, that treatments should be directed to improving the functions of affected brain cell-types.

Adverse effects
The choice of a medication is partly determined by Its safety profile, particularly for serious adverse events (SAEs), less so for minor adverse events (AE). Following are summaries of reports of SAEs for the above drugs.
Clemastine: no SAEs seen in several clinical trials. Dantrolene: other than caused by preexisting disease, SAEs were rare [237]. Erythropoietin: only 3 of 50 studies attributed SAEs to erythropoietin but those used very high-dosages (> 200,000 IU) [238]. If chosen for use in patients with BD, MDD, or SCZ, erythropoietin should be low dosage, 10,00 IU given IM once monthly.
Fluoxetine: Beasely et al. obtained data from 25 double-blind clinical trials involving 4016 patients with MDD randomized to treatment that included fluoxetine 20 to 80 mg/d [240]. At a dose of 20 mg/d, fluoxetine-treated patients had a discontinuation rate due to adverse events that was not statistically significantly different from that in placebo recipients. Nevertheless, although in this article the suggested dose is only 10 mg/d in combination with lithium, caution is necessary because the possibility exists of inducing mania, even though that is reported mainly with doses of ≥ 20 mg/day [241].
Lithium: SAEs, include hypercalcemia, hypothyroidism [242], nephrogenic diabetes insipidus, and renal insufficiency [243]. Low dosage (< 100 mgs daily) benefited Alzheimer's dementia; thus 75 mg daily might be an appropriate dose in BD, MDD, SCZ, AD or PTSD. It should not be used if glomerular filtration rate is < 60 mL/min. Memantine: no significantly increased SAEs [244]. Minocycline: a review showed six patients with a hypersensitivity syndrome, six with serum sickness, and 24 with drug-induced lupus [245]. Minocycline has been used in thousands of persons over manyyears, so SAEs are rare. ANA test should be negative before prescribing it.
Piracetam: a large study, did not show more SAEs than with placebo [246]. Riluzole: nine studies did not show increased SAEs [247].

The need for clinical trials
Clinical trials would demonstrate validity of both the proposed concept and the choice of drugs for treatment; and would also show the percentage of patients benefitting from the chosen combination of drugs administered for a particular psychiatric condition. In a clinical trial that randomly assigns participants to receive active drugs or placebo, the primary objective would be to show ≥ 30% better results from active drugs than from placebo. In order to provide deeper coverage and lower dosages, two drugs rather than one should be used.

Discussion
The proposed model is a basis for formulating treatment of major neuro-psychiatric conditions. The model overcomes the problem of addressing the many genetic, mitochondrial, and metabolic impairments associated with the discussed conditions; it proposes drugs that address the dominant combination of brain cell-types, which simplifies use of known data; future availability of data that involve subtypes of brain cell-types, will allow improvement of the model. Drugs are discussed which address the brain's five cell-types that affect the occurrence of BD, MDD, SCZ, AD, and PTSD.
It is acknowledged that a single treatment combination would not benefit all patients with a particular psychiatric phenotype, because an identical clinical phenotype may result from several, different combinations of cell-types, only one of which is dominant compared with the other as compared with the others; thus, the suggested treatments would benefit only those subjects whose condition is caused by a particular set of dominant cell-types. Nevertheless, benefit to even 20-30% of patients would be clinically useful. It is, also, worth noting that in studies of all of the mentioned conditions, 'conflicting' results are reported; yet because an identical clinical phenotype may incorporate different combinations of cell-types, it is possible that the different results might not be actually 'conflicting' but correct.
The determination of the dominant change in the brain cell-type in any particular condition is based upon the findings in published reports and, thus, the diligence with which the literature is reviewed. Data for the present article were derived primarily from citations provided by PubMed and Google Scholar. Citations found in key articles are often very relevant; for example, ref [60] cited an article that reviewed studies in 1415 patients with SCZ [248]. For the four psychiatric conditions used to illustrate the proposed approach, the 61 citations included seven that reviewed 1136 separate, published reports [21,23,37,41,56,60,68], and one that reviewed 15 publicly available data sets [37]. Reviews generally refer to very large numbers of subjects; e.g., one "review of reviews" analyzed 683 reports in 26 reviews involving 55,561 patients [65].
It is relevant to ask, 'which is primary, alterations in brain cell-types or the changed behavior?' because if it is the altered behavior that causes the changes in brain cell-types, then addressing the changes in those cells would not affect behavior. However, and fortunately for the thesis of this article, a number of studies show that the altered brain cells precede the altered behavior. In one such study, Santarelli et al. noted that chronic antidepressant treatments increase adult hippocampal neurogenesis; they found that destruction by X-irradiation of the region of mouse brain containing the hippocampus, prevented both the neurogenesis and the behavioral benefits of two classes of antidepressants, showing that stimulation of neurogenesis in the hippocampus underpins the behavioral effects of chronic antidepressants [249]. Related to that, the induction of depression-like behavior in chronic stress models was associated with elimination of postsynaptic dendritic spines and a loss of PFC projection neurons [250]; and ablation of astrocytes in the PFC, which deprives neurons of trophic support, increased anhedonia, anxiety, and helplessness, similar to the depression-like behaviors caused by chronic unpredictable stress [251]. The above studies may be interpreted as demonstrating that. for the relationship between neurons and behavior, it is the neuronal action that precedes a behavioral change, not vice versa.
There are, of course, many caveats to consider: (1) Cell-type families have sub-types; therefore, designating a drug as addressing a generic cell-type may be inaccurate. However, there are scores of subtypes. E.g., Tasic et al. found 56 glutamatergic and 61 GABAergic, cortical neuronal subtypes [252], and Bickoff et al. documented multiple subtypes of spinal, inhibitory interneurons [253]. Currently, data are sparse concerning the presence and distribution of cell-subtypes in the major psychiatric conditions, and there is no information regarding a possible, differential effect of medication upon subtypes. In future, when more information about them becomes available, it will be possible to establish if subtypes must be separately accounted for. (2) One report found no consistent evidence of astrocytic impairments in 11 psychiatric disorders; that study determined cell-type by using single cell transcriptomic data integrated with genome wide association study results [254]. However, there is variability of gene expression by astrocytes, e.g., Stahlberg et al. identified two subpopulations of astrocytes with distinct gene expression profiles [255]. Thus, absence of either a specific gene or set of genes would incorrectly label some cells as being 'not astrocytes' . Further, glial fibrillary acidic protein (GFAP) is the cardinal cytological marker of astrocytes but GFAP-negative astrocytes also exist, e.g., Tatsumi et al. saw that brain regions such as the basal forebrain, thalamic nuclei, and deep cerebellar nuclei, tended to lack GFAP-positive astrocytes [256]; thus also in this case, the absence of GFAP would incorrectly identify cells as being 'not astrocytes' . (3) Another limitation of the proposed algorithm is that it assumes uniform changes of cell-types across all brain regions. That this is not necessarily the case, is exemplified by observations made by Gandal et al. in brains from patients with autistic spectrum disorder (ASD) [257]. They saw a regional gradients, with three-fold to four-fold more differentially expressed genes in cell-types (particularly excitatory neurons, interneurons, and oligodendrocytes) in occipital and parietal cortices than in the PFC.
Despite these theoretical limitations there is, nevertheless, theoretical consistency in that the proposed target of therapy is the changed direction of number or function of the individual cell-types as seen in individual psychiatric conditions. The accuracy of the cell-type(s) as target can be tested by clinical trial since knowing the changed direction of cell-types provides clinical benefit if a specific drug or combination of drugs corrects both the changed direction and the clinical condition. 1. For most psychiatric conditions, pathogenesis depends on impairment of either number or function among the five families of cell-types in the brain. Cell-types are storehouses of and, therefore, surrogates for, pathogenically important genetic, metabolic, and environmental inputs. The dominantly changed cell-types in a given condition permits algorithmic formulation of a pharmacotherapeutic regimen. 2. Because an identical clinical phenotype may result from different combinations and permutations of impaired celltypes, no single combination of cell-types accounts for all cases; in a given set of subjects one can only discern the combination that is dominant as compared with the others; therefore, different studies may produce discrepant results, all of which could be accurate. 3. Eleven drugs are presented that benefit either all or some of the five cell-types and from among which treatment may be formulated for each condition. 4. Clinical trial is required to validate both the proposed concept and choice of drugs.
Data availability Data regarding the suggested drugs are available from the author.

Competing interests The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.