Viruses undergo natural selection just like anything else. They don't make proteins that do error-checking when copying their DNA or RNA (coronavirus is an RNA virus), so they get a lot more mutations than humans do. A lot of those mutations are duds, some are successful. The successful ones will churn out tons more. But any virus that makes it into an RBC will be a dud because it can't replicate there. If it had a mutation that was actually favorable to selecting an RBC, that would be the end of its line. That's what I meant about being selected against.
The article you linked is interesting. It took a little ways into it before I realized that their entire study was computer simulations. They don't have the virus itself and didn't manufacture any viral proteins to fully test their idea. =/
An attack on heme should necessitate invasion of the cell, since almost all hemoglobin is intracellular. The orf proteins aren't structural proteins of the virion. They're only made once its RNA gets into the cell and gets translated. However, mature red blood cells don't have ribosomes, so no orf proteins could be produced. Reticulocytes are the last stage in making red blood cells (1-2% of what's in circulation) and do still have some ribosomes left, but they're in the process of getting rid of them. I can't find any references to COVID-19 and reticulocyte count.
The article suggests that free-floating hemoglobin is attacked by viral proteins after it is "infected," as though the hemoglobin itself were a cell. The E2 glycoprotein and envelope protein are structural proteins and present on the virion. Let's say it did attack free hemoglobin - this wouldn't affect the hemoglobin contained within the red cells and so it wouldn't affect the oxygen carrying capacity.
I think that article gets pretty dubious once it gets to the discussion section. Hemoglobin in men is higher than women, but the hemoglobin lab value refers to hemoglobin contained within the red cells, not free floating. Then it says it's more common in older patients, but a lot of the elderly are anemic with low hemoglobin. Diabetics have higher glycated hemoglobin, but glycated hemoglobin is expressed as a percentage and is not representative of a higher total hemoglobin. Free heme is toxic and the body wouldn't just leave it floating around waiting for the virus.
There is a different way to interfere with blood oxygen transport, which is to alter the configuration of the hemoglobin molecule. The body does it physiologically - high affinity shape in the lungs for picking up oxygen, low affinity shape in the tissues for dropping off some oxygen (only about 25% is delivered normally). With carbon monoxide poisoning, hemoglobin is locked into a high affinity state and has trouble dropping off oxygen at the tissues. If it's locked in a low affinity shape, it will have trouble picking up oxygen. Methemoglobin would have a low affinity. Maybe the oxygenation problem is physiologic?
If you have IOS, there is an app "CD Antigens" where you can look up these markers and where they're found.
I don't think reticulocytes would be subject to much attack; I don't remember at what stage they are released from the marrow and get into the bloodstream.
Testing the idea in this case would be very difficult, given the stark differences between in vivo and in vitro.
The body has so *many* interdependencies and feedback loops.
Also, according to MedCram they're seeing not low saturation, but low PO2; but that didn't specify what stage of the disease progress, early or late.