How to dynamic switch sqlalchemy model through graphene relay?

Question

In our system, we have two similar(but not same) databases. So I built these sqlalchemy models:

# base.py
Base = declarative_base()

class T1(Base):
    __tablename__ = 't1'
    id = Column(Integer, primary_key=True)
    name = Column(String)

# production1.py
from . import base

class T1(base.T1):
    status1 = Column(String)

# production2.py
from . import base

class T1(base.T1):
    status2 = Column(String)

# sessions.py
engine1 = create_engine(**production1_params)
session1 = scoped_session(sessionmaker(bind=engine1))
engine2 = create_engine(**production2_params)
session2 = scoped_session(sessionmaker(bind=engine2))

Then I can access different database by:

import production1, production2
session1().query(production1.T1)
session2().query(production2.T2)

Now, I want to build our API system by graphql. First, I inherit from SQLAlchemyConnectionField to support database switching.

class SwitchableConnectionField(SQLAlchemyConnectionField):
    def __init__(self, type, *args, **kwargs):
        kwargs.setdefault('db_type', String())
        super

    @classmethod
    def get_query(self, model, info, sort=None, **args):
        session = get_query(args['db_type'])
        query = session.query(model)
        ...

But when I want to define my nodes, I found the definitions must be:

import production1, production2

class Production1Node(SQLAlchemyObjectType):
    class Meta:
        model = production1,T1
        interfaces = (Node,)

class Production2Node(SQLAlchemyObjectType):
    class Meta:
        model = production2.T1
        interfaces = (Node,)

There are two nodes definitions to support different databases. But I want to do something like:

import base

class ProductionNode(SQLAlchemyObjectType):
    class Meta:
        model = base.T1
        interfaces = (Node,)

So that I can switch similar model at run time. However, even though I try to inherit from Node, I can't implement it. Does anyone know what should I do?