I can't say authoritatively, but the alignments (the multi-row protein letter alignment) looks like output from something like clustal, e.g. the clustalomega software https://www.ebi.ac.uk/Tools/msa/clustalo/ domains and arrow markup may have been done using an image editing tool as an after-step using manual inspection of coordinates and/or domain sequences (just guessing here though, maybe there is a way to automatically mark this up)
Did you link to the correct figure? This one? If you did, that is just an annotated gene sequence with the amino acids shown. The box around the domain was added, I suspect. If you aren't making a lot of these, you can use use your favorite sequence/text editor (NOTE MICROSOFT WORD) and a vector graphics program to draw this. You can get the boundaries from the site you describe even if you cannot download the image. By doing it in a graphics program you are sure to get high enough resolution for easy publishing down the road.
The cooler kids may be using some software for at least use in the lab.
Unipro UGENE - an example along the lines if what you show with it is here. I've been moving towards using that to replace SerialCloner that I used for a long time, but has not been kept as up-to-date in recent years.
Lots of people like ApE - an example along the lines if what you show with it is here.
See more suggestions from a couple years ago here.
It spells out clearly at the bottom the site hasn't been updated in over six years.
RasMol is featured. While Rasmol was an important step in bringing structure visualization to non-structural biologists, it has long been superseded by more powerful molecular visualization packages, such as PyMOL and UCSF Chimera, or more modern/web-oriented tools that allowed easy use & sharing, such as Jmol and others in that vein.
In this case, the few things you are mining from there are probably fine; however, it is always best to be wary & make sure you are using up-to-date information.
The Conserved Domain Database (CDD) is kept up-to-date and ProViz allows generating some nice visuals or exploring for cases when you are just interested in the Protein coding sequence.
It is probably a bit overkill, but I would give the latex package texshade a try. It is a shader for alignments of both nucleotides and peptides and produces quality publication figures. You provide your alignment (or even single sequence) as a text file and you then add all the aesthetics you need.
I can't say authoritatively, but the alignments (the multi-row protein letter alignment) looks like output from something like clustal, e.g. the clustalomega software https://www.ebi.ac.uk/Tools/msa/clustalo/ domains and arrow markup may have been done using an image editing tool as an after-step using manual inspection of coordinates and/or domain sequences (just guessing here though, maybe there is a way to automatically mark this up)
I'll throw in
seqvisrinto the hat. Might be a bit of a pain for this use case though (but should probably still work).