Hello, and thanks for your help!
Because the level shifters are not multiples of a standard cell height, according to the Cadence documentation they must be placed in their own power domains. This was done, I believe, correctly, but still did not align them with the rows.
The tape-out has passed now, so I gave up trying to find a solution to this and instead concentrated on getting the cells in the right place through less elegant methods. I was able to get limited success by using the adjustPowerDomainToAlignRows command, which resized the various power domains in order to align their rows.
This worked for all but one level-shifter power domain, where the limited number of cells allowed me to place them manually.
Once aligned, I set them all to fixed, so that they were not moved during later stages of the flow.
Lastly, the followpins were in each case not created correctly. In order to solve this I wrote a tcl script that identifies the level shifter rows and creates the wire via the addWire command.
In this way I was able to create all the wires with minimal effort.
So unfortunately I was never able to identify the cause of the problem, but was able to get around it with the methods outlined above.
Thank you for your help in any case, it is really greatly appreciated!