An MDMA analogue like 5-MAPB or 4-MMC with low-dose selegiline seems to be the combination with the best MDMA-likeness to neurotoxicity ratio. (targets the two main mechanisms (if hyperthermia is physically avoided) - toxic metabolites and serotonergic neurotoxicity caused by simultaneous dopamine release). Taking selegiline should have the additional effect of a longer/stronger dopaminergic effect due to slowed dopamine degradation. 
Vitamin C, vitamin E, ALA, and agmatine likely provide further neuroprotection. Antipsychotics like clozapine are an effective measure for MDMA-overdose-caused hyperthermia (only as an emergency measure; otherwise it would dull the effects significantly).

Selegiline is a drug prescribed for Parkinson's disease and depression (for which the patch form - EMSAM - is used in the US).

Mechanism
In a 1995 study, it was shown using selegiline and the SSRI fluoxetine, that damage to neuronal membranes from MDMA stems largely from the uptake of dopamine by serotonin transporters, thus causing dopamine to accumulate in serotonergic neurons, where it is broken down mainly by MAO-B (whereas elsewhere, it's metabolised mostly by MAO-A). This deamination by MAO-B creates hydrogen peroxide, which is claimed to be responsible for much of MDMA's neurotoxicity.
 

Selegiline pharmacology
Selegiline is an MAOi; it inhibits the enzymes that metabolise monoamines, such as dopamine, serotonin, norepinephrine, and trace amines, increasing their intersynaptic concentration.
Selegiline, along with safinamide and rasagiline, is selective for the MAO-B subtype at certain dosages. The MAO-B enzyme metabolises mostly beta-phenethylamine and dopamine, though it has been found, that MAO-A is mostly responsible for dopamine breakdown, however, it seems likely that it is indeed MAO-B which metabolises dopamine in serotonergic nerve terminals. In addition, there seems to be evidence that MAO-B metabolises serotonin in serotonergic neurons as well, also producing hydrogen peroxide.
Selegiline, as well as the beta-phenethylamine it increases (the "endogenous amphetamine"), are agonists of TAAR1. This is also a target of amphetamine, probably being responsible for part of its action. TAAR1 agonism may be responsible for selegiline's catecholaminergic activity enhancer (CAE) activity.^[1]
In addition, selegiline (especially orally) is metabolised into the levorotary forms of amphetamine and methamphetamine, which act as norepinephrine and dopamine releasing agents, though much weaker in terms of dopamine release than their dextrorotary "conventional" counterparts (detroamphetamine and dextromethamphetamine).^[2] 
Selegiline's CAE effect, amphetamine metabolites, and decreased dopamine metabolism suggest it might enhance MDMA's effect, even in the absence of functional MAO-A inhibition, though it seems from the above-linked studies, that for rodents no significant additional hyperthermia or head-twitch response is observed. This might be due to the fact that most dopamine is metabolised by MAO-A in blood platelets, rather than MAO-B in serotonergic neurons, and thus only the small fraction of dopamine breakdown, that is responsible for neurotoxic effects, is targeted by selegiline administration. 
Still, it is a potentially risky practice to add selegiline to MDMA use, with possible individual variation in drug metabolism causing MAO-A inhibition at dosages that usually only inhibit MAO-B. Therefore, low dosages of <10 mg, maybe 5 mg, are more reasonable.
 

Evidence
Aside from the Sprague 1995 study linked above, which showed that selegiline (also called Deprenyl) alleviated markers of neurotoxicity in MDMA-exposed rats, in vitro culture had significantly reduced free radical formation when administered selegiline before MDMA, and MAO-B deficient rats where shown to have no serotonin depletion from MDMA (indicating lacking damage to serotonergic neurons - the main targets of MDMA neurotoxicity). Another study from 2007 showed MAO-B dependent damage to mitochondria (including their DNA), alleviated by selegiline.
Notably, I didn't find any study testing the combination in humans, nor any anecdotal experience report.

Selectivity/dosage:
At dosages of <10 mg/day (orally), selegiline is selective for MAO-B, thus not posing a significant risk of serotonin syndrome^[3] (BUT maybe your body works differently and is highly sensitive). If you want to try this potentially dangerous combination, ideally microdose the selegiline at dosages below 5 mg. 
Patches (EMSAM) might theoretically be useful due to a lower amount of metabolites (reported 70% reduction), but they are risky due to their much higher bioavailability, up to 50-fold, and thus simultaneous MAO-A inhibition at normal dosages (which is the goal of the patches, intended as antidepressants).^[4] In addition, it releases the drug slowly over the course of a day, making it less predictable when planning a specific dosage in combination with MDMA.

Which pharmacological aspects of MDMA do we look for?
Besides the dopamine and norepinephrine release and reuptake inhibition, which causes stimulation and euphoria, the specifics of MDMA's effects are likely due to interaction with serotonin receptors. However, it is not entirely clear how serotonin release leads to prosocial, enactogenic effects. Experiments with receptor antagonists have shown that the psychedelic receptor 5-HT2A (which modulates dopamine and prolactin release) is relevant for this effect, as well as 5-HT1B and 5-HT1A, which has downstream effects on increased oxytocin release.^[5] Notably, it not only indirectly stimulates but also directly binds to, as a partial agonist, to these receptors, in addition to 5-HT2B and the sigma receptors, which might be responsible for some of its effects. It also binds to alpha-adrenergic receptors, possibly contributing to its anxiolytic effects.^[6] 
Hormonally, MDMA has been shown to elevate the neuropeptides oxytocin and vasopressin, as well as the steroids DHEA and cortisol. The increase in oxytocin seems to be correlated with prosocial effects, and the increase in DHEA with euphoric ones^[6] (maybe one should try to supplement DHEA or intranasal oxytocin?). The hormonal effects are, however, much less studied in MDMA analogues than in MDMA itself, so I'll focus on the activity as a triple monoamine releaser and serotonergic receptor activity instead.

Analogues
The four following MDMA analogues belong into the classes of benzofurans (5-MAPB and 6-APB) and cathinones (4-MMC and 3-MMC), meaning their cores are different from that of MDMA, and thus form different metabolites.
All of them are triple monoamine releasing agents and (except for 5-MAPB) their reuptake inhibitor - leading to increased extracellular concentration of serotonin, dopamine and norepinephrine (noradrenaline). At the same time, all have affinity to the above-mentioned serotonin receptors, mostly as partial agonists (just as MDMA). 
Importantly, they have none of the metabolites of MDMA, such as alpha-methyldopamine, HHA or HHMA, which are neurotoxic through their auto-oxidation into quinones and their binding to gluthanione, forming neurotoxic thioether conjugates.^[7] Therefore, unless there are yet-to-be-discovered neurotoxic metabolites of these drugs, their damage must come primarily from hyperthermia, dopamine breakdown by MAO-B, or RNS formation, which can be alleviated with the methods described in the section below.

5-MAPB
It is metabolised into 5-APB (which is a very similar MDMA analogue, however with stronger 5-HT2A agonist and thus likely stronger psychedelic effects), which is subsequently metabolised into 3-carboxymethyl-4-hydroxy amphetamine, and another product is 3-carboxymethyl-4-hydroxy methamphetamine.^[8] No catechols or MDA are formed, and so the conversion to quinones or thioether conjugates doesn't happen. However, 5-APB is a potent agonist of the 5-HT2B receptor^[9], which might lead to cardiotoxicity with regular use^[10] (not recommened for neurotoxicity and monoamine depletion reasons anyway).
It has a halflife of 6.5 hours.^[11]
In terms of effects, the potent 5-HT1B agonism^[12] might make this a particularly empathogenic/pro-social compound.^[13] However, the lack of direct serotonin 1A receptor activity might make it less oxytocin-releasing, though this mighe be alleviated by stronger serotonin releasing activity (but lower NE activity), and thus indirect 5-HT1A agonism^[14], as well as the metabolite 5-APB's activity at the receptor. The lower norepinephrine-releasing activity makes this compound likely less stimulating.
5-MAPB has been found to create hyperthermia (and hypertension, tremor and convulsions) in humans^[15], though this is likely to be a case of an overdose. In rat liver cells, 5-MAPB and it's metabolite 5-APB have been shown to cause cell death (cytotoxicity), greater than MDMA.^[16]

In humans, I found one report of fatal intoxication, though it has been in combination with several other compounds.^[18]
In rodents, 5-MAPB caused similar serotonin depletion as MDMA.^[19]

However, much of the toxicity of 5-MAPB compared to MDMA might be caused by use of the same amount of each compound in these studies, even though 5-MAPB is a cca. 3x stronger monoamine releaser than MDMA at the same dose^[20][21]. This might simply mean a way too high dosage has been used, leading to more serotonergic hyperthermia + the SERT-mediated toxic dopamine breakdown by MAO-B. Thus at a 3 times lower dosage, it might be a safer alternative to MDMA, especially in combination with low-dose selegiline and hypothermic compounds such as agmatine, alleviating most of the remaining neurotoxicity potential.

6-APB
It is much stronger than MDMA - 12x more potent than at the dopamine transporter, 6.5x stronger at the noradrenaline transporter, and 2.4x stronger at the serotonin transporter.^[22] With this altered ratio of monoamine release, it can be expected to be more akin to a stimulant like methamphetamine in it's effects. In terms of total monoamine increase measured, all tested benzofurans were about 3x more potent than MDA (so about 9x more potent than MDMA^[23]), and 6-APB has been shown to be the most potent benzofuran in terms of dopamine release.^[24]
In addition, 6-APB was found to bind with high affinity to alpha-adrenergic receptors (similar to MDMA - potentially calming), to 5-HT2A receptors (psychedelic), 5-HT1A receptors (oxytocin-mediating), and, most strongly, to 5-HT2B receptors, which poses potential cardiotoxic risks.^[25] The alpha-adrenergic and 5-HT2A agonism make 6-APB a potentially more hyperthermia-inducing drug (which can lead to significant damage, but can also relatively simply be avoided by staying in a cool environment).
The effects begin within 1-2 hours and last for about 7 hours.^[26]
It does not form quinone or thioether metabolites, which contribute to MDMA neurotoxicity (the main 6-APB metabolite was 4-carboxymethyl-3-hydroxy amphetamine).^[27] 
No cytotoxic effects have been found in one cell culture study^[28]. There exists a report of 6-APB-caused psychosis, though this has been in combination with cannabis.^[29]
Overall, 6-APB seems like a more stimulant and psychedelic analogue, with little data on neurotoxicity, though by its similarity to 5-APB it can be assumed to have similar cytotoxic oxidative effects, but less so (it has been found to be less toxic to liver cells than 5-APB, the active 5-MAPB metabolite)^[30]

4-MMC aka mephedrone
Mephedrone a triple monoamine reuptake inhibitor and releasing agent (as is MDMA, though mephedrone is more of a reuptake inhibitor).^[31] It is also a strong 5-HT2A agonist, suggesting it might have psychedelic properties (though strangely, it is not a proper hallucinogen). The lack of direct actiity on the 5-HT1A receptor might mean lower oxytocin release. 
It is a relatively short-acting drug, with effects beginning after 15 minutes, and lasting 2-3 hours (when taken orally). 
^[32]
Mephedrone is commonly insufflated, and is reported to have effects similar to MDMA.^[31]

It's metabolites are mostly nor-mephedrone, which is psychoactive itself, as a stimulant (DAT and NET inhibition with less SERT inhibition), DHMMC (which has a similar but weaker profile), and several mostly inactive metabolites like 5-hydroxy-mephedrone.^[33] Again, no quinones or thioethers are produced, and none of the studied metabolites has been shown to have neurotoxic properties.
Interestingly the article "Clinical Pharmacology of the Synthetic Cathinone Mephedrone"^[34] from 2017 reports:

Regarding the possible long-term toxicity of mephedrone, the fact that the drug possesses structural and pharmacological similarities to MDMA, amphetamines, and cathinone suggests the likelihood that repeated and/or prolonged use produces similar consequences on neurochemical and neuropsychological function. From the limited results to date, it should be pointed out that repeated mephedrone administration in experimental animals has not shown evidence of neurotoxicity to monoaminergic systems in the brain [42, 88–91^[35]].

One study reports decreased serotonin transporter function in rats administered 4-MMC, but the rats were purposefully kept in a high-temperature environment.^[36]
Mephedrone induces hyperthermia^[37] and potentiates the neurotoxicity of methamphetamine and MDMA, but does not itself cause dopaminergic neurotoxicity. This has lead to the conclusion that mephedrone functions atypically at the dopamine transporter^[38] (which might possibly be the reason behind its relative non-neurotoxicity).
One rat study showed oxidative damage to rat neurons as well as dopamine receptor downregulation.^[39]
As opposed to MDMA, mephedrone has not been shown to cause microglial activation, thus the pathway leading to RNS damage is likely nonexistent for mephedrone.^[40] Cognitive damage (working memory worsening) has been found in mice after "binge-treatment" of mephedrone.^[41] There have been some deaths due to mephedrone overdoses.^[42]

Overall, mephedrone seems like a surprisingly safer MDMA alternative, if hyperthermia is avoided (many studies showing harm in rodents used elevated ambient temperature). The working memory deficits shown in rats are concerning, but likely a consequence of high dosages and/or hyperthermia.

3-MMC
3-MMC aka metaphedrone is commonly insufflated
3-MMC inhibits the serotonin transporter much less than mephedrone or MDMA, while significantly inhibiting DAT and NET, suggesting a more stimulant, rather than enactogenic effect.^[43] However, the 5-HT1 agonism may lead to oxytocin release, leading to empathogenic effects, confirmed by users.^[44] It binds strongly to alpha-adrenergic receptors, which might pose vasoconstriction risk, but the lack of 5-HT2A agonism makes the risk of hyperthermia lower.^[43] It is capable of producing hyperthermia, it lasts for around an hourwhen insufflated, and is reported to be a weaker version of mephedrone or MDMA in terms of its effects. The main metabolites are 3-methylephedrine and 3-methylnorephedrine, with no known neurotoxic effect.^[45] 3-MMC has been shown to create ROS (and RNS), and damage liver cells.^[46] Inhibition of the enzyme CYP2D6 has been shown to be protective, suggesting that genetic variations in the expression of this enzyme may affect the toxicity of 3-MMC use, with "extensive and ultrarapid metabolisers" experiencing significantly more toxicity^[45] (which is likely true of MDMA and its other analogues too).
Two deaths due to 3-MMC (likely in combination with other drugs) have been reported in Sweden^[47], 5 severe poisonings in the Netherlands^[48], and adeath following pure 3-MMC exposure in France.^[49]
Overall, 3-MMC appears to be an inferior alternative to 4-MMC, having a short halflife, shown oxidative stress toxicity, and potentially neurotoxic metabolites.